CN112818679A

CN112818679A - Event type determination method and device and electronic equipment

Info

Publication number: CN112818679A
Application number: CN201911122660.6A
Authority: CN
Inventors: 刘英箎; 李泉志; 张琼; 司罗
Original assignee: Alibaba Group Holding Ltd
Current assignee: Alibaba Group Holding Ltd
Priority date: 2019-11-15
Filing date: 2019-11-15
Publication date: 2021-05-18

Abstract

The application discloses an event category determination method and device, a traffic information broadcasting system, method and device and electronic equipment. The event type determination method comprises the following steps: determining words included in the event text to be classified and event element types of the words; extracting a sub-network through event features included by an event classification model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element vector of the event element type; and determining the event category of the event text according to the event characteristics through an event category prediction sub-network included in the event classification model. By adopting the processing mode, event category identification is carried out by combining the semantic information of the event text and the event element type information of the event text; therefore, the accuracy of event category identification can be effectively improved.

Description

Event type determination method and device and electronic equipment

Technical Field

The application relates to the technical field of natural language processing, in particular to an event category determining method and device, an event element type prediction model establishing method and device, an event classification model establishing method and device, a traffic information broadcasting system, method and device, a device failure occurrence mode determining method and device, an event occurrence mode determining method and device, an enterprise acquisition event occurrence mode determining method and device, a traffic accident occurrence mode determining method and device, an event occurrence mode determining system, method and device, an event element determining system, method and device and electronic equipment.

Background

The event text classification technology is a research hotspot in the field of natural language, and can automatically classify and process the text.

The process of a typical event text classification method is as follows. Firstly, acquiring text data of a plurality of labeled event types to form a training data set; then, learning from the training data set through a machine learning algorithm to obtain an event type prediction model; and finally, taking the text to be detected as input data of an event type prediction model, and automatically identifying the event type of the text through the model.

However, in the process of implementing the present invention, the inventor finds that the prior art solution has at least the following problems: because each word in the event text is simply embedded as a word, the event category to which the text belongs is determined purely based on information on a semantic level, and the event element information of the event text is ignored, the accuracy of event category identification is low.

Disclosure of Invention

The application provides an event category determining method, which aims to solve the problem of low accuracy of event category identification in the prior art. The application further provides an event category determining device, an event element type prediction model establishing method and device, an event classification model establishing method and device, a traffic information broadcasting system, a traffic information broadcasting method and device, a device failure occurrence mode determining method and device, an event occurrence mode determining method and device, an enterprise acquisition event occurrence mode determining method and device, a traffic accident occurrence mode determining method and device, an event occurrence mode determining system, method and device, an event element determining system, method and device and electronic equipment.

The application provides a method for determining an equipment failure occurrence mode, which comprises the following steps:

acquiring a text related to equipment failure;

determining words included in the text and equipment fault element types of the words;

extracting a sub-network through the equipment fault feature included in the fault element determination model, and extracting equipment fault features including text semantic information and equipment fault element type information according to a word vector of a word and an element type vector of the equipment fault element type;

determining equipment fault element information according to the equipment fault characteristics through a fault element prediction sub-network included in a fault element determination model;

and determining equipment fault occurrence mode information according to the equipment fault element information.

Optionally, the device fault element type is determined by the following steps:

aiming at each word, taking the word as a target word, and determining a text segment comprising the target word and adjacent words thereof;

acquiring word vectors of a target word and adjacent words; determining word position vectors of the target word and the adjacent words according to word position information of the adjacent words in the text segment;

and aiming at each target word, determining the equipment fault element type of the target word according to the word vector of the target word and adjacent words thereof and the word position vector by an equipment fault element type prediction model.

Optionally, the method further includes:

and learning to obtain the equipment fault element type prediction model from the corresponding relation between the equipment fault text for training and the equipment fault element type marking information of the words included in the equipment fault text for training.

Optionally, the extracting a device fault feature sub-network included in the fault element determination model, according to the word vector of the word and the element type vector of the device fault element type, extracts a device fault feature including text semantic information and device fault element type information, and includes:

determining word vectors of words including equipment fault element type information according to the word vectors of the words and the element type vectors;

and extracting the equipment fault characteristics through the equipment fault characteristic extraction sub-network according to the word vector comprising the equipment fault element type information.

Optionally, the method further includes:

and learning to obtain the fault element determination model from the corresponding relation between the training equipment fault text and the equipment fault element marking information.

Optionally, the method further includes:

and determining relevant parameters of equipment operation according to the equipment fault occurrence mode information.

The present application further provides an apparatus for determining an equipment failure occurrence mode, including:

the fault text acquisition unit is used for acquiring a device fault related text;

the element type determining unit is used for determining words included in the text and equipment fault element types of the words;

a fault feature extraction unit, which is used for extracting a sub-network according to the equipment fault feature included in the fault element determination model, and extracting the equipment fault feature including text semantic information and equipment fault element type information according to the word vector of the word and the element type vector of the equipment fault element type;

a failure element determination unit configured to determine device failure element information based on the device failure characteristics, using a failure element prediction sub-network included in a failure element determination model;

and the failure mode determining unit is used for determining the failure occurrence mode information of the equipment according to the failure element information of the equipment.

The application provides an electronic device, including:

a processor; and

a memory for storing a program for implementing a method for determining a failure occurrence mode of a device, the device being powered on and the program for performing the method by the processor, and performing the steps of: acquiring a text related to equipment failure; determining words included in the text and equipment fault element types of the words; extracting a sub-network through the equipment fault feature included in the fault element determination model, and extracting equipment fault features including text semantic information and equipment fault element type information according to a word vector of a word and an element type vector of the equipment fault element type; determining equipment fault element information according to the equipment fault characteristics through a fault element prediction sub-network included in a fault element determination model; and determining equipment fault occurrence mode information according to the equipment fault element information.

The application provides an event occurrence mode determining method, which comprises the following steps:

acquiring an event related text;

determining words included in the text and event element types of the words;

extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type;

determining event element information according to the event characteristics through an event element prediction sub-network included in an event element determination model;

and determining event occurrence mode information according to the event element information.

Optionally, the event occurrence mode information includes: patterns of event occurrences related to public transportation facilities; the method further comprises the following steps:

and determining the dispatching information of the public transport personnel according to the event occurrence mode information.

The application provides an event occurrence mode determination device, including:

an event text acquisition unit for acquiring an event-related text;

the element type determining unit is used for determining words included in the text and event element types of the words;

an event feature extraction unit, which is used for extracting sub-networks according to the event features included in the event element determination model, and extracting event features including text semantic information and event element type information according to the word vector of the word and the element type vector of the event element type;

an event element specifying unit configured to specify event element information based on the event feature by using an event element prediction sub-network included in an event element specifying model;

and the event occurrence mode determining unit is used for determining the event occurrence mode information according to the event element information.

The application provides an electronic device, including:

a processor; and

a memory for storing a program for implementing the event occurrence pattern determination method, the apparatus performing the following steps after being powered on and running the program of the method by the processor: acquiring an event related text; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information according to the event characteristics through an event element prediction sub-network included in an event element determination model; and determining event occurrence mode information according to the event element information.

The application provides a method for determining an occurrence mode of an enterprise acquisition event, which comprises the following steps:

acquiring a text related to an enterprise purchase event;

determining words included in the text and the type of the enterprise purchase event elements of the words;

extracting a sub-network of enterprise acquisition event features included by an enterprise acquisition event element determination model, and extracting enterprise acquisition event features including text semantic information and enterprise acquisition event element type information according to a word vector of a word and an element type vector of an enterprise acquisition event element type;

determining enterprise acquisition event element information according to the characteristics of the enterprise acquisition event by an enterprise acquisition event element prediction sub-network included in an enterprise acquisition event element determination model;

and determining the occurrence mode information of the enterprise acquisition event according to the element information of the enterprise acquisition event.

The application provides a business acquisition event occurrence pattern determining device, includes:

the event text acquisition unit is used for acquiring the text related to the enterprise purchase event;

the element type determining unit is used for determining words included in the text and enterprise purchase event element types of the words;

the event feature extraction unit is used for extracting sub-networks from the enterprise acquisition event features included in the enterprise acquisition event element determination model, and extracting enterprise acquisition event features including text semantic information and enterprise acquisition event element type information according to the word vector of the word and the element type vector of the enterprise acquisition event element type;

the event element determining unit is used for determining enterprise acquisition event element information according to the characteristics of the enterprise acquisition events through an enterprise acquisition event element prediction sub-network included in an enterprise acquisition event element determining model;

and the event occurrence mode determining unit is used for determining the enterprise acquisition event occurrence mode information according to the enterprise acquisition event element information.

The application provides an electronic device, including:

a processor; and

a memory for storing a program for implementing the method for determining the occurrence mode of the enterprise acquisition event, wherein after the device is powered on and the program for the method is run by the processor, the following steps are executed: acquiring a text related to an enterprise purchase event; determining words included in the text and the type of the enterprise purchase event elements of the words; extracting a sub-network of enterprise acquisition event features included by an enterprise acquisition event element determination model, and extracting enterprise acquisition event features including text semantic information and enterprise acquisition event element type information according to a word vector of a word and an element type vector of an enterprise acquisition event element type; determining enterprise acquisition event element information according to the characteristics of the enterprise acquisition event by an enterprise acquisition event element prediction sub-network included in an enterprise acquisition event element determination model; and determining the occurrence mode information of the enterprise acquisition event according to the element information of the enterprise acquisition event.

The application also provides an event category determination method, which comprises the following steps:

determining words included in the event text to be classified and event element types of the words;

extracting a sub-network through event features included by an event classification model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element vector of the event element type;

and determining the event category of the event text according to the event characteristics through an event category prediction sub-network included in the event classification model.

Optionally, the event element type is determined by the following steps:

and aiming at each target word, determining the event element type of the target word according to the word vector of the target word and adjacent words thereof and the word position vector by an event element type prediction model.

Optionally, the extracting, by the event feature extraction sub-network included in the event classification model, the event feature including text semantic information and event element type information according to the word vector of the word and the element vector of the event element type includes:

determining a word vector of a word including element type information according to the word vector of the word and the element vector of the event element type;

and extracting the event features according to the word vector comprising the element type information through the event feature extraction sub-network.

The present application also provides an event category determination device, including:

the element type determining unit is used for determining words included in the event text to be classified and event element types of the words;

an event feature extraction unit, which is used for extracting sub-networks according to the event features included in the event classification model, and extracting event features including text semantic information and event element type information according to the word vectors of the words and the element vectors of the event element types;

and the event type determining unit is used for determining the event type of the event text according to the event characteristics by using the event type prediction sub-network included in the event classification model.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the event class determination method, the device being powered on and the program for implementing the method being executed by the processor to perform the steps of: determining words included in the event text to be classified and event element types of the words; extracting a sub-network through event features included by an event classification model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element vector of the event element type; and determining the event category of the event text according to the event characteristics through an event category prediction sub-network included in the event classification model.

The application also provides a method for constructing an event element type prediction model, which comprises the following steps:

acquiring a training sample set; the training samples comprise training event texts, words included in the training event texts and corresponding relations among event element type marking information of the words;

constructing a network structure of the model; wherein the network structure comprises a feature extraction sub-network and an event element type prediction sub-network; the feature extraction sub-network is used for extracting event features used for predicting event element types according to word vectors of words of the event texts; the event element type prediction sub-network is used for acquiring a predicted value of an event element type of a word according to the event characteristics;

and training the network parameters of the model according to the training sample set to obtain an event element type prediction model.

Optionally, the feature extraction sub-network is specifically configured to extract the event features according to word vectors of words of the event text and adjacent words thereof, and word position vectors, where the event features include context semantic information of the words.

The present application also provides an event element type prediction model construction device, including:

the data acquisition unit is used for acquiring a training sample set; the training samples comprise training event texts, words included in the training event texts and corresponding relations among event element type marking information of the words;

the network construction unit is used for constructing a network structure of the model; wherein the network structure comprises a feature extraction sub-network and an event element type prediction sub-network; the feature extraction sub-network is used for extracting event features used for predicting event element types according to word vectors of words of the event texts; the event element type prediction sub-network is used for acquiring a predicted value of an event element type of a word according to the event characteristics;

and the network training unit is used for training the network parameters of the model according to the training sample set to obtain an event element type prediction model.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the event element type prediction model construction method, the apparatus performing the following steps after being powered on and running the program of the method by the processor: acquiring a training sample set; the training sample comprises a training event text, words included in the training event text and corresponding relations among event element type marking information of the words; constructing a network structure of the model; wherein the network structure comprises a feature extraction sub-network and an event element type prediction sub-network; the feature extraction sub-network is used for extracting event features used for predicting event element types according to word vectors of words of the event texts; the event element type prediction sub-network is used for acquiring a predicted value of an event element type of a word according to the event characteristics; and training the network parameters of the model according to the training sample set to obtain an event element type prediction model.

The application also provides an event classification model construction method, which comprises the following steps:

acquiring a training sample set; the training sample comprises corresponding relations between training event texts and event category marking information;

constructing a network structure of the model; wherein the network structure comprises an event feature extraction sub-network and an event category prediction sub-network; the feature extraction sub-network is used for extracting event features according to the word vectors of all words of the event text and the element vectors of the event element types with the words as target words, wherein the features comprise text semantic information and event element type information; the event type prediction sub-network is used for acquiring a predicted value of an event type of an event text according to the event characteristics;

and training the network parameters of the model according to the training sample set to obtain an event classification model.

The present application further provides an event classification model building apparatus, including:

the data acquisition unit is used for acquiring a training sample set; the training sample comprises corresponding relations between training event texts and event category marking information;

the network construction unit is used for constructing a network structure of the model; wherein the network structure comprises an event feature extraction sub-network and an event category prediction sub-network; the feature extraction sub-network is used for extracting event features according to the word vectors of all words of the event text and the element vectors of the event element types with the words as target words, wherein the features comprise text semantic information and event element type information; the event type prediction sub-network is used for acquiring a predicted value of an event type of an event text according to the event characteristics;

and the network training unit is used for training the network parameters of the model according to the training sample set to obtain an event classification model.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the event classification model construction method, wherein the following steps are executed after the device is powered on and the program for implementing the method is run by the processor: acquiring a training sample set; the training sample comprises corresponding relations between training event texts and event category marking information; constructing a network structure of the model; wherein the network structure comprises an event feature extraction sub-network and an event category prediction sub-network; the feature extraction sub-network is used for extracting event features according to the word vectors of all words of the event text and the element vectors of the event element types with the words as target words, wherein the features comprise text semantic information and event element type information; the event type prediction sub-network is used for acquiring a predicted value of an event type of an event text according to the event characteristics; and training the network parameters of the model according to the training sample set to obtain an event classification model.

The application also provides a traffic accident occurrence mode determining method, which comprises the following steps:

acquiring a text related to a traffic accident;

determining words included in the text and traffic accident element types of the words;

extracting traffic accident features including text semantic information and traffic accident element type information according to a word vector of a word and an element type vector of a traffic accident element type through a traffic accident feature extraction sub-network included in a traffic accident element determination model;

determining traffic accident factor information according to the traffic accident characteristics through a traffic accident factor prediction sub-network included in a traffic accident factor determination model;

and determining traffic accident occurrence mode information according to the traffic accident factor information.

Optionally, the element information includes: traffic accident grade information; the method further comprises the following steps:

and determining the quantity of the traffic accidents of each traffic accident grade according to the traffic accident grade information.

The present application further provides a traffic accident occurrence pattern determining apparatus, including:

the accident file acquisition unit is used for acquiring a text related to the traffic accident;

the element type determining unit is used for determining words included in the text and traffic accident element types of the words;

the accident feature determination unit is used for extracting traffic accident features including text semantic information and traffic accident element type information according to the word vector of the words and the element type vector of the traffic accident element types through the traffic accident feature extraction sub-network included in the traffic accident element determination model;

an accident element determination unit for determining traffic accident element information based on the traffic accident characteristics by using a traffic accident element prediction sub-network included in a traffic accident element determination model;

and the accident mode determining unit is used for determining the traffic accident occurrence mode information according to the traffic accident factor information.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the traffic accident occurrence pattern determination method, the apparatus performing the following steps after being powered on and running the program of the method through the processor: acquiring a text related to a traffic accident; determining words included in the text and traffic accident element types of the words; extracting traffic accident features including text semantic information and traffic accident element type information according to a word vector of a word and an element type vector of a traffic accident element type through a traffic accident feature extraction sub-network included in a traffic accident element determination model; determining traffic accident factor information according to the traffic accident characteristics through a traffic accident factor prediction sub-network included in a traffic accident factor determination model; and determining traffic accident occurrence mode information according to the traffic accident factor information.

The present application further provides a traffic information broadcasting system, including:

the first service end is used for acquiring a text related to the traffic accident; determining words included in the text and traffic accident element types of the words; extracting traffic accident features including text semantic information and traffic accident element type information according to a word vector of a word and an element type vector of a traffic accident element type through a traffic accident feature extraction sub-network included in a traffic accident element determination model; determining traffic accident factor information according to the traffic accident characteristics through a traffic accident factor prediction sub-network included in a traffic accident factor determination model; determining traffic accident occurrence mode information according to the traffic accident element information;

the second server is used for determining accident multi-occurrence section information and accident occurrence condition information according to the traffic accident occurrence mode information; determining the position information of a vehicle to be early-warned; if the distance between the vehicle position and the accident multi-occurrence section is smaller than the distance threshold value and the accident occurrence condition is established, determining traffic accident early warning information, and sending the early warning information to a client associated with the vehicle to be early warned;

and the client is used for receiving the early warning information so as to adjust the driving mode of the vehicle according to the early warning information.

The application also provides a traffic information broadcasting method, which comprises the following steps:

determining accident multi-occurrence section information and accident occurrence condition information according to the traffic accident occurrence mode information;

determining the position information of a vehicle to be early-warned;

if the distance between the vehicle position and the accident multi-occurrence section is smaller than the distance threshold value and the accident occurrence condition is established, determining traffic accident early warning information;

and sending the early warning information to a client associated with the vehicle to be early warned.

and receiving traffic accident early warning information sent by the server so as to adjust the driving mode of the vehicle according to the early warning information.

The application also provides a traffic information reports device, includes:

the accident related information determining unit is used for determining accident multi-occurrence section information and accident occurrence condition information according to the traffic accident occurrence mode information;

the vehicle position determining unit is used for determining the position information of the vehicle to be early-warned;

the early warning information determining unit is used for determining traffic accident early warning information if the distance between the vehicle position and the accident multi-occurrence section is smaller than a distance threshold value and the accident occurrence condition is satisfied;

and the early warning information sending unit is used for sending the early warning information to a client side associated with the vehicle to be early warned.

The application also provides a traffic information reports device, includes:

and the information receiving unit is used for receiving the traffic accident early warning information sent by the server so as to adjust the vehicle running mode according to the early warning information.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the traffic information broadcasting method, wherein the device executes the following steps after being powered on and running the program of the method through the processor: determining accident multi-occurrence section information and accident occurrence condition information according to the traffic accident occurrence mode information; determining the position information of a vehicle to be early-warned; if the distance between the vehicle position and the accident multi-occurrence section is smaller than the distance threshold value and the accident occurrence condition is established, determining traffic accident early warning information; and sending the early warning information to a client associated with the vehicle to be early warned.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the traffic information broadcasting method, wherein the device executes the following steps after being powered on and running the program of the method through the processor: and receiving traffic accident early warning information sent by the server so as to adjust the driving mode of the vehicle according to the early warning information.

The present application also provides an event occurrence pattern determination system, including:

the server is used for acquiring at least one event text; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information; receiving an event occurrence mode information acquisition request aiming at the target event element, which is sent by a client; determining event occurrence pattern information including the target event element; returning event occurrence pattern information including the target event element to the client;

and the client is used for sending the request to the server and receiving the event occurrence mode information comprising the target event element.

The application also provides an event occurrence mode determining method, which comprises the following steps:

acquiring at least one event text; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information;

receiving an event occurrence mode information acquisition request aiming at a target event element, which is sent by a client;

determining event occurrence pattern information including the target event element;

and sending back the event occurrence mode information comprising the target event element to the client.

sending an event occurrence mode information acquisition request aiming at the target event element to a server;

and receiving the event occurrence mode information including the target event element returned by the server.

The present application also provides an event occurrence mode determining apparatus, including:

the event occurrence mode generating unit is used for acquiring at least one event text; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information;

the request receiving unit is used for receiving an event occurrence mode information acquisition request aiming at a target event element, which is sent by a client;

an event occurrence mode query unit, configured to determine event occurrence mode information including the target event element;

a mode information returning unit configured to return event occurrence mode information including the target event element to the client.

the request sending unit is used for sending an event occurrence mode information acquisition request aiming at the target event element to the server;

and the information receiving unit is used for receiving the event occurrence mode information including the target event element returned by the server.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the event occurrence pattern determination method, the apparatus performing the following steps after being powered on and running the program of the method by the processor: at least one event text is obtained in advance; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information; receiving an event occurrence mode information acquisition request aiming at a target event element, which is sent by a client; determining event occurrence pattern information including the target event element; and sending back the event occurrence mode information comprising the target event element to the client.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the event occurrence pattern determination method, the apparatus performing the following steps after being powered on and running the program of the method by the processor: sending an event occurrence mode information acquisition request aiming at the target event element to a server; and receiving the event occurrence mode information including the target event element returned by the server.

The present application also provides an event element determination system, including:

the server is used for receiving an event element information determination request aiming at a target event text sent by the client; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; returning event factor information to the client;

and the client is used for sending the request to the server and receiving the event element information.

The application also provides an event element determination method, which comprises the following steps:

receiving an event element information determination request aiming at a target event text sent by a client;

determining words included in the text and event element types of the words;

determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model;

and returning the event factor information to the client.

sending an event element information determination request aiming at a target event text to a server;

and receiving the event factor information returned by the server.

The present application also provides an event element determination device, including:

the request receiving unit is used for receiving an event element information determination request aiming at a target event text sent by a client;

an event feature determination unit, configured to extract, according to an event feature extraction sub-network included in the event element determination model, an event feature including text semantic information and event element type information based on a word vector of a word and an element type vector of the event element type;

an event element determination unit configured to determine event element information of the event text based on the event feature by using an event element prediction sub-network included in an event element determination model;

and the information returning unit is used for returning the event factor information to the client.

the request sending unit is used for sending an event element information determination request aiming at the target event text to the server;

and the information receiving unit is used for receiving the event element information returned by the server.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the event element determination method, the device performing the following steps after being powered on and the program for the method being executed by the processor: receiving an event element information determination request aiming at a target event text sent by a client; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; and returning the event factor information to the client.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the event element determination method, the device performing the following steps after being powered on and the program for the method being executed by the processor: sending an event element information determination request aiming at a target event text to a server; and receiving the event factor information returned by the server.

The present application also provides a computer-readable storage medium having stored therein instructions, which when run on a computer, cause the computer to perform the various methods described above.

The present application also provides a computer program product comprising instructions which, when run on a computer, cause the computer to perform the various methods described above.

Compared with the prior art, the method has the following advantages:

the event category determining method provided by the embodiment of the application determines words and event element types of the words included in an event text to be classified; extracting a sub-network through event features included by an event classification model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element vector of the event element type; determining the event type of the event text according to the event characteristics through an event type prediction sub-network included in an event classification model; the processing mode enables event type identification to be carried out by combining semantic information of the event text and event element type information of the event text; therefore, the accuracy of event category identification can be effectively improved.

According to the method for constructing the event element type prediction model, a training sample set is obtained; the training samples comprise training event texts, words included in the training event texts and corresponding relations among event element type marking information of the words; constructing a network structure of the model; wherein the network structure comprises a feature extraction sub-network and an event element type prediction sub-network; the feature extraction sub-network is used for extracting event features used for predicting event element types according to word vectors of words of the event texts; the event element type prediction sub-network is used for acquiring a predicted value of an event element type of a word according to the event characteristics; training the network parameters of the model according to the training sample set to obtain an event element type prediction model; the processing mode can combine the word body semantic information of the event text words and the information of the peripheral text to construct an event element type prediction model; therefore, the prediction accuracy of the event element type prediction model can be effectively improved.

According to the method for constructing the event classification model, a training sample set is obtained; the training sample comprises corresponding relations between training event texts and event category marking information; constructing a network structure of the model; wherein the network structure comprises an event feature extraction sub-network and an event category prediction sub-network; the feature extraction sub-network is used for extracting event features according to the word vectors of all words of the event text and the element vectors of the event element types with the words as target words, wherein the features comprise text semantic information and event element type information; the event type prediction sub-network is used for acquiring a predicted value of an event type of an event text according to the event characteristics; training the network parameters of the model according to the training sample set to obtain an event classification model; the processing mode can combine the word body semantic information and the event element type information of the event text words to construct an event classification prediction model; therefore, the prediction accuracy of the event classification model can be effectively improved.

According to the method for determining the equipment fault occurrence mode, the relevant text of the equipment fault is obtained; determining words included in the text and equipment fault element types of the words; extracting a sub-network through the equipment fault feature included in the fault element determination model, and extracting equipment fault features including text semantic information and equipment fault element type information according to a word vector of a word and an element type vector of the equipment fault element type; determining equipment fault element information according to the equipment fault characteristics through a fault element prediction sub-network included in a fault element determination model; determining equipment fault occurrence mode information according to the equipment fault element information; the processing mode can automatically determine the equipment failure occurrence mode, help to find the situation with multiple equipment failures and take preventive measures; therefore, the determination efficiency of the equipment failure occurrence mode can be effectively improved.

According to the method for determining the event occurrence mode, the relevant text of the event is obtained; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information; the processing mode can automatically extract elements from the event text, classify the mode condition of the case, automatically determine the event occurrence mode, help to determine the frequent situation of the event and take preventive measures; therefore, the determination efficiency of the event occurrence mode can be effectively improved.

According to the method for determining the equipment failure occurrence mode, the relevant text of the enterprise purchase event is obtained; determining words included in the text and the type of the enterprise purchase event elements of the words; extracting a sub-network of enterprise acquisition event features included by an enterprise acquisition event element determination model, and extracting enterprise acquisition event features including text semantic information and enterprise acquisition event element type information according to a word vector of a word and an element type vector of an enterprise acquisition event element type; determining enterprise acquisition event element information according to the characteristics of the enterprise acquisition event by an enterprise acquisition event element prediction sub-network included in an enterprise acquisition event element determination model; determining the occurrence mode information of the enterprise acquisition event according to the element information of the enterprise acquisition event; the processing mode enables the elements of the enterprise co-purchasing event to be automatically extracted, classifies the mode condition of the co-purchasing event, automatically determines the occurrence mode of the co-purchasing event and helps to determine the situation of multiple co-purchasing events so as to specify the enterprise co-purchasing plan; therefore, the determining efficiency of the enterprise co-purchase event occurrence mode can be effectively improved.

According to the traffic accident occurrence mode determining method provided by the embodiment of the application, the text related to the traffic accident is obtained; determining words included in the text and traffic accident element types of the words; extracting traffic accident features including text semantic information and traffic accident element type information according to a word vector of a word and an element type vector of a traffic accident element type through a traffic accident feature extraction sub-network included in a traffic accident element determination model; determining traffic accident factor information according to the traffic accident characteristics through a traffic accident factor prediction sub-network included in a traffic accident factor determination model; determining traffic accident occurrence mode information according to the traffic accident element information; the processing mode can automatically extract elements of the traffic accident case of the traffic management department, classify the mode situation of the accident, automatically determine the traffic accident occurrence mode, help to determine the situation of frequent traffic accidents, and take preventive measures; therefore, the determining efficiency and the determining accuracy of the traffic accident occurrence mode can be effectively improved.

The traffic information broadcasting system provided by the embodiment of the application acquires the text related to the traffic accident through the first server; determining words included in the text and traffic accident element types of the words; extracting traffic accident features including text semantic information and traffic accident element type information according to a word vector of a word and an element type vector of a traffic accident element type through a traffic accident feature extraction sub-network included in a traffic accident element determination model; determining traffic accident factor information according to the traffic accident characteristics through a traffic accident factor prediction sub-network included in a traffic accident factor determination model; determining traffic accident occurrence mode information according to the traffic accident element information; determining accident multi-occurrence section information and accident occurrence condition information according to the traffic accident occurrence mode information through a second server; determining the position information of a vehicle to be early-warned; if the distance between the vehicle position and the accident multi-occurrence section is smaller than the distance threshold value and the accident occurrence condition is established, determining traffic accident early warning information, and sending the early warning information to a client associated with the vehicle to be early warned; the client receives the early warning information so as to adjust the driving mode of the vehicle according to the early warning information; the processing mode can remind the driver of paying attention to safe driving in time; therefore, driving safety can be effectively improved.

The event occurrence mode determining system provided by the embodiment of the application obtains at least one event text through the server; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information; receiving an event occurrence mode information acquisition request aiming at the target event element, which is sent by a client; determining event occurrence pattern information including the target event element; returning event occurrence pattern information including the target event element to the client; this processing manner makes it possible to determine event occurrence pattern information including a target event element; therefore, the efficiency and the accuracy of determining the event occurrence mode can be effectively improved.

The event element determining system provided by the embodiment of the application receives an event element information determining request aiming at a target event text, which is sent by a client through a server; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; returning event factor information to the client; the processing mode enables the event element information of the target event text to be determined; therefore, the efficiency and the accuracy of determining the event elements can be effectively improved.

Drawings

FIG. 1 is a flow chart of an embodiment of an event category determination method provided herein;

FIG. 2 is a detailed flowchart of an embodiment of an event category determination method provided in the present application;

FIG. 3 is a schematic diagram of a model structure of an embodiment of an event category determination method provided in the present application;

FIG. 4a is a schematic diagram of word vector embedding of an embodiment of an event category determination method provided by the present application;

FIG. 4b is a schematic diagram of word position vector embedding of an embodiment of an event category determination method provided by the present application;

FIG. 5 is a schematic diagram of event element type vector embedding of an embodiment of an event category determination method provided in the present application;

FIG. 6 is a detailed flowchart of an embodiment of an event category determination method provided in the present application;

FIG. 7 is a schematic diagram of an embodiment of an event category determining apparatus provided herein;

FIG. 8 is a schematic diagram of an embodiment of an electronic device provided herein;

FIG. 9 is a flow chart of an embodiment of a method for determining a failure mode of a device provided herein;

FIG. 10 is a schematic diagram of an embodiment of an apparatus failure mode determination device provided herein;

FIG. 11 is a schematic diagram of an embodiment of an electronic device provided herein;

FIG. 12 is a detailed flow chart of an embodiment of an event occurrence pattern determination method provided herein;

FIG. 13 is a detailed schematic diagram of an embodiment of an event occurrence pattern determination apparatus provided herein;

FIG. 14 is a schematic diagram of an embodiment of an electronic device provided herein;

FIG. 15 is a flowchart illustrating an embodiment of a method for determining an occurrence pattern of an enterprise acquisition event provided herein;

FIG. 16 is a detailed schematic diagram of an embodiment of an enterprise acquisition event occurrence pattern determining apparatus provided by the present application;

FIG. 17 is a schematic diagram of an embodiment of an electronic device provided herein;

FIG. 18 is a flow diagram of an embodiment of a method for constructing an event element type prediction model provided herein;

FIG. 19 is a diagram of an embodiment of an event element type prediction model construction device provided in the present application;

FIG. 20 is a schematic diagram of an embodiment of an electronic device provided herein;

FIG. 21 is a flowchart of an embodiment of a method for constructing an event classification model provided herein;

FIG. 22 is a schematic view of an embodiment of an event classification device provided herein;

FIG. 23 is a schematic diagram of an embodiment of an electronic device provided herein;

FIG. 24 is a flow chart of an embodiment of a traffic accident pattern determination method provided herein;

FIG. 25 is a schematic view of an embodiment of a traffic accident occurrence pattern determination apparatus provided herein;

FIG. 26 is a schematic view of an embodiment of an electronic device provided herein;

fig. 27 is a schematic view of an embodiment of a traffic information broadcasting system provided in the present application;

fig. 28 is a device interaction diagram of an embodiment of the traffic information reporting system provided by the present application;

FIG. 29 is a schematic diagram of an embodiment of an event occurrence pattern determination system provided herein;

FIG. 30 is a device interaction diagram of an embodiment of an event occurrence pattern determination system provided herein;

FIG. 31 is a schematic diagram of an embodiment of an event element determination system provided herein;

FIG. 32 is a device interaction diagram of an embodiment of an event element determination system provided herein.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. This application is capable of implementation in many different ways than those herein set forth and of similar import by those skilled in the art without departing from the spirit of this application and is therefore not limited to the specific implementations disclosed below.

In the application, an event type determination method and device, an event element type prediction model construction method and device, an event classification model construction method and device, a traffic information broadcasting system, method and device, an equipment failure occurrence mode determination method and device, an event occurrence mode determination method and device, an enterprise acquisition event occurrence mode determination method and device, a traffic accident occurrence mode determination method and device, an event occurrence mode determination system, method and device, an event element determination system, method and device, and electronic equipment are provided. Each of the schemes is described in detail in the following examples.

First embodiment

Please refer to fig. 1, which is a flowchart illustrating an embodiment of a method for determining an event category according to the present application, wherein an executing entity of the method includes an event category determining apparatus. The event category determination method provided by the application comprises the following steps:

step S101: determining words included in the event text to be classified and event element types of the words.

The event text to be classified describes an event and is text related to the event. For example, the event text may be a case of an illegal case, a notice of a car company merger, a utility company equipment failure report, and the like.

For example, the text of the event to be classified is a complaint book, the description of the complaint book is 1, I cooperate with friends for 300 ten thousand projects, after the oral engagement project is finished, the two parties distribute 50% of the profit of the project on average, but no written cooperation agreement is signed between the two parties. 2. I are responsible for project earlier-stage work, including (business, bid, technology, product initial-stage type selection, delivery, matching construction units, signing and closing and the like) partner responsible for project work, including (early-stage investment, business part after winning bid, product purchase and payment) 3. the project pays 30 ten thousand more costs due to her reasons. "

In the embodiment, a user specifies an event text to be classified through a client, and the client sends an event classification request aiming at the event text to a server; and the server receives the request, analyzes the request through the deployed event type determining device, and obtains the event text to be classified.

The event text to be classified is composed of a plurality of words, and can be texts in various languages, such as Chinese or English. When the event text is a Chinese text, determining each word included in the event text to be classified through a Chinese word segmentation algorithm, for example, after the text is 'spoken to agree on a project, the two parties equally distribute profits of the project', and the word segmentation result includes the following words: oral, appointment, project, end, back, etc. When the event text is an English text, the word segmentation result comprises English words appearing in the text.

In specific implementation, the existing word segmentation algorithm can be adopted to perform word segmentation processing on the event text. The existing word segmentation algorithm can be divided into three categories: a word segmentation method based on character string matching, a word segmentation method based on understanding and a word segmentation method based on statistics. Whether the method is combined with the part-of-speech tagging process or not can be divided into a simple word segmentation method and an integrated method combining word segmentation and tagging. The word segmentation algorithm belongs to the mature prior art, and is not repeated herein, and any one of the existing word segmentation algorithms can be selected according to actual requirements.

After the word segmentation processing is performed, the event element type to which the word belongs can be determined for each word in each event text. The event element refers to an element constituting an event, and although the content and the space of the event text are different, each event text cannot lack any element. The event element type may also be referred to as an event element dimension or an event element name, and accordingly, the event element may be referred to as an event element value. If the event text is regarded as unstructured information, the event elements are structured information of the event text, and the structured information can be information which is interesting to the user.

For example, the element types of case text include: time, place, victim, illegal event, etc., wherein a specific time period, type of incident, sex, age bracket of victim, type of illegal event (such as robbery, theft, personal injury, fraud), etc., are the essential information of the event. According to the event element information, an event occurrence mode can be determined, such as high probability of bus theft in the daytime, certain type of failure of the equipment A of the power company in drought, and the like.

Please refer to fig. 2, which is a flowchart illustrating an embodiment of a method for determining an event category according to the present application. In this embodiment, the event element type may be determined by the following steps:

step S201: and aiming at each word, taking the word as a target word, and determining a text segment comprising the target word and adjacent words thereof.

With this embodiment, each word included in the event text can be expanded into a sequence of words using the context. In specific implementation, each word included in the event text can be used as a target word, and a preset number of n adjacent words are respectively selected forwards and backwards (left and right sides) by taking the target word as a center. For example, the target word w _0 is expanded into the following text sequence: [ w _ -n, w _ -n +1, … w _ -1, w _0, w _1, …, w _ n ]. Wherein, w _0 is the target word, w _ -n +1, … w _ -1 are the adjacent n words on the left side of the target word, and w _1, …, w _ n are the adjacent n words on the right side of the target word. From the number of words point of view, the text segment may be a text string with a fixed number of words (1+2 n).

For example, the text for the case scenario is "… afternoon girl touched chest on bus …", where the text segment for the target word "bus" may be [ afternoon, girl, on, bus, up, touched, chest ], assuming n is 2, "bus" is W0, "girl" is W-2, "on" is W-1, "up" is W1, "touched" is W2.

It should be noted that, when a target word is used as a center and is expanded to the left and right sides to form a text string (i.e., a text fragment) with a preset number of words, the same number of words can be expanded to the left and right sides, for example, 5 words are expanded to the left and right sides; or expanding different numbers of words to the left and right, such as expanding 3 words to the left and expanding 5 words to the right.

In specific implementation, if the target word is the initial word of the event text and n words expanded leftward cannot be obtained, n "empty words" may be used to represent the word centered on the target word and expanded leftward.

Step S203: acquiring word vectors of a target word and adjacent words; and determining word position vectors of the target word and the adjacent words according to the word position information of the adjacent words in the text fragment.

After the target word and the adjacent word are obtained in step S201, the process may proceed to this step, and vector embedding (embedding) is performed on the position information of the target word, the adjacent word, and the adjacent word in the text segment, respectively, to obtain vector expression forms of various information. And for the target word and the adjacent words, organizing the text segments according to word vectors in a word embedding mode, and acquiring the word vectors capable of expressing semantics so as to conveniently mine the features of the words for determining the element types according to the word vectors.

In one example, the method provided in the embodiment of the present application further includes the following steps: a dictionary is constructed, wherein the dictionary may include all words that appear in the event text. In specific implementation, a dictionary can be constructed by scanning all training articles. After the dictionary is constructed, the words in the dictionary may also be indexed such that each word corresponds to a unique numerical identifier. In addition, after the dictionary is built, the word vector training processing can be performed on the words in the dictionary, and the word vector (word embedded) corresponding to the words in the dictionary can be obtained through training by using a word vector training algorithm (word2vec, such as fastText) by using all training data. In specific implementation, the trained word vector can be directly used.

The word vector embedding method and the word vector embedding device can perform word vector embedding on the target word and the adjacent words through the dictionary inquiring process to obtain the word vectors of the target word and the adjacent words. As shown in fig. 4a, in a specific implementation, a word vector (word embedding) corresponding to each word may be obtained by querying a word vector matrix (word embedding matrix) according to the unique identifier of each word. Taking an example where the lexicon includes 10000 words, the word vector matrix is a 10000 word vector dimension (custom) matrix. Each row of the matrix corresponds to a word vector. This matrix may be a randomly initialized matrix that may be updated by training the word vectors.

In order to make the event element type prediction model (especially in the CNN network) better know which word in the text segment needs to be subjected to element type judgment and distinguish text sequences generated by two adjacent words in the original text, in this embodiment, each word in the text segment is numbered in sequence, which is called as a word position. Each word corresponds to a word position, and the word position information comprises position information of the words in the text passage. These positions may be input into the event element type prediction model through a position embedding (position embedding) method.

Vector representations of various word positions (such as 0, -1, 1 and the like) can be obtained by training in the training process of the event element type prediction model.

For convenience of description, in the embodiments of the present application, the word position information corresponding to the text segment is expressed as: p {... P_t-1,p_t,p_t+1.., wherein p_tIs a target word w_tWord position in text segment S, p_t-1Is the 1 st word w adjacent to the left side of the target word_t-1Word position in text segment S, p_t+1Is the 1 st word w adjacent to the right side of the target word_t+1Lexeme in text segment SAnd (4) placing.

The word position information may be determined according to the position of the word in the text segment with respect to the target word, for example, if the word length of the text segment S is fixed to 11, then P { -5, -4, … 0 …,4,5 }; it may also be determined directly from the position of the respective word in the text passage, as indicated by p ═ {0, … 10 }.

As shown in fig. 4b, in this embodiment, a position vector (position embedding) corresponding to each word may be obtained by querying a position vector matrix (position embedding) according to the word position number of each word. Taking the number of words in the text segment as 11 as an example, the position vector matrix is a matrix of 11-word position vector dimension (self-defined). Each row of the matrix corresponds to a word position. This matrix may be a randomly initialized matrix that may be updated by training the word position vector.

For example, when the text segment S { "empty word", "just", "oath", "present", "managed academy", "institution length", and P { -5, -4, … 0 …,4,5}, the following words and word position vectors are obtained: the word position vector corresponding to the position number-5 of the 1 st "empty word", the word position vector corresponding to the position number-4 of the 2 nd "empty word", the word position vector corresponding to the position number-3 of the 3 rd "empty word", the word position vector corresponding to the position number-2 of the 4 th "empty word", the word position vector corresponding to the position number-1 of the 5 th "empty word", the word position vector corresponding to the position number 0 of the "immediately" position vector corresponding to the "immediately" position number 0 ", the word position vector corresponding to the position number 1 of the" solemk ", the word position vector corresponding to the position number 2 of the" just "position number 3 of the" position vector corresponding to the "position number 2 of the" right ", the word position vector corresponding to the position number 4 of the" department of charge ", and the word position vector corresponding to the position number 5 of the" courtyard ".

After the word vectors and the word position vectors of the target word and the adjacent words thereof are obtained, the next step can be entered.

Step S205: and aiming at each target word, determining the event element type of the target word according to the word vector of the target word and adjacent words thereof and the word position vector by an event element type prediction model.

Please refer to fig. 3, which includes a schematic diagram of a network structure of the event element type prediction model according to the present embodiment. In this embodiment, for each target word, a vector sequence formed by connecting a Word Embedding (WE) and a word position vector (PE) corresponding to each word in a text segment corresponding to the target word is input to the event element type prediction model. The model includes a feature extraction sub-network and an event element type prediction sub-network. Through the feature extraction sub-network, the features influencing the element types of the target words can be extracted from the limited text range around the target words. Inputting the representation into an event element type prediction sub-network, obtaining the scores of the target word on each element type through a full connection layer, normalizing the scores through a softmax function, and obtaining the type with the highest score as the element type of the target word.

The feature extraction sub-network may adopt a variety of deep neural network structures, including but not limited to: convolutional Neural Networks (CNNs), bi-directional long-short term memory networks (bilstms), and so on. The event element type prediction sub-network may include a Fully connected layer (full connected layer) and a normalization layer (softmax function).

According to the method provided by the embodiment of the application, the characteristic extraction sub-network based on the Bi-LSTM is adopted, so that the long-distance dependence relationship between words can be modeled, modeling can be performed from two directions, and the long-distance dependence relationship between the words can determine the characteristic which influences the element type of the target word; therefore, the accuracy of the feature which influences the element type of the target word can be effectively improved.

It should be noted that the same target word may not have the same event element type in different event text contexts. For example, the event element type of the event text "hit the" Chinese word "home" by a person at home "is" place element ", and the event element type of the event text" hit the "Chinese word" home "by a person on the way to go home while sitting on a bus may be" unknown ". In this embodiment, through the processing in the above steps 1 to 3, different event element types of the same word in different text contexts can be identified, and the main principle is as follows: the words are expanded into a text sequence by using the context, so that the event element types can be identified according to the context semantics of the words. By adopting the processing mode, the identification accuracy of the event element types can be effectively improved, so that the event classification accuracy is improved.

In concrete implementation, the event element type of each word may be determined in other manners than the above-described embodiment. For example, the words are not expanded into text sequences, and the event element types of the words are directly determined according to word ontology semantics; or, although the words are expanded into text sequences, the event element types of the words are determined only according to the word ontology semantics and the context semantics of the words without considering the word position factors, and the like.

Step S103: and extracting the event features comprising text semantic information and event element type information according to the word vector of the word and the element vector of the event element type through the event feature extraction sub-network included in the event classification model.

The inventor of the invention finds that the event element type can play an auxiliary role in judging the event type, so that the method provided by the embodiment of the application combines the word vector of each word in the event text with the event element type to expand the information quantity predicted by the event type, thereby effectively improving the accuracy of the event characteristics influencing the event type and further improving the accuracy of event classification.

Fig. 3 is a schematic diagram of a network structure of the event classification model according to this embodiment, where the model includes an event feature extraction sub-network and an event category prediction sub-network. The event feature extraction sub-network can be used for extracting event features used for predicting event categories according to word vectors of all words of an event text and element vectors of event element types with the words as target words, and the features comprise text semantic information and event element type information.

In specific implementation, a word vector WE of each word in the event text and a vector (KE) of the event element type may be concatenated to form a vector with the length of the sum of the lengths of WE and KE, the event text is converted into a sequence (for example, W0 to W5 in fig. 2) formed by the vectors, the vector sequence is input into an event feature extraction sub-network, and the event feature including text semantic information and event element type information is extracted through the sub-network.

Wherein, the vector representation of various event element types (such as place, time, victim and the like) can be in the training process of the event classification model. In the present embodiment, the element vector of the event element type is a 10-dimensional vector. When the event classification model is trained, vector representation of event element types can be initialized firstly, the vector representation tends to be stable gradually along with the gradual reduction of loss values in the training process, and when the model training is finished, the final vector representation can be stored as model parameters so as to carry out embedding processing of the element types according to data in the prediction stage.

As shown in fig. 5, in this embodiment, an element type vector corresponding to each element type can be obtained by querying the element type vector matrix according to the number of each element type. Taking the number of element types as 5 as an example, the element type vector matrix is a 5 × element type vector dimension (custom) matrix. Each row of the matrix corresponds to one element type. This matrix may be a randomly initialized matrix that may be updated by training the event classification model.

Step S105: and determining the event category of the event text according to the event characteristics through an event category prediction sub-network included in the event classification model.

And inputting the event characteristics extracted in the previous step into an event category prediction sub-network, and determining the event category of the event text through the sub-network. The event characteristics are obtained according to the ontology semantic information of each word in the event text and the event element type information of the word, and the event element type information is related to the dimension of the event to be classified, so that the event classification effect is improved.

The network structure of the feature extraction sub-network included in the event classification model can be a convolutional neural network or a bidirectional long-short term memory network, and the like. The event category prediction subnetwork may include a fully connected layer and a normalization layer. The event classification model comprises a feature extraction sub-network and an event category prediction sub-network, and the feature extraction sub-network and the event category prediction sub-network can have the same structure as the corresponding sub-networks in the event element type prediction model, and have the differences of different input data, different feature properties and different output data.

It should be noted that, to implement the method provided in the embodiment of the present application, an event element type prediction model and an event classification model are first constructed. The event element type prediction model can be obtained by learning from first training data. The event classification model may be learned from second training data.

Please refer to fig. 6, which is a flowchart illustrating an embodiment of a method for determining an event category according to the present application. In this embodiment, the method further includes the steps of:

step S601: and learning the event element type prediction model from a first training sample set.

The first training sample set includes a plurality of first training samples. The first training sample comprises training event texts, words included in the training event texts and corresponding relations among event element type marking information of the words. Table 1 shows a first training sample set of the present embodiment.

TABLE 1 first training sample set

After a first training sample set is obtained, the event element type prediction model can be obtained by learning from the first training sample set through a deep learning algorithm. Since the deep learning algorithm belongs to the mature prior art, it is not described herein again.

When the event element type prediction model is trained, the loss function of the event element type prediction sub-network can adopt a binary cross entropy function or a common cross entropy function.

Step S603: and learning the event classification model from a second training sample set.

The second training sample set includes a plurality of second training samples. The second training sample comprises corresponding relations between training event texts and event category marking information. Table 2 shows a second training sample set of the present embodiment.

TABLE 2 second training sample set

In this embodiment, for each element type, an event classification model corresponding to the element type is trained. And determining event categories of a plurality of element dimensions through event classification models of different element dimensions according to the event characteristics.

In specific implementation, a multi-classification event classification model can be adopted, and event classes of multiple element dimensions of an event file can be predicted and obtained sequentially through one event classification model.

As can be seen from the foregoing embodiments, the event category determining method provided in the embodiments of the present application determines words included in an event text to be classified and event element types of the words; extracting a sub-network through event features included by an event classification model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element vector of the event element type; determining the event type of the event text according to the event characteristics through an event type prediction sub-network included in an event classification model; the processing mode enables event type identification to be carried out by combining semantic information of the event text and event element type information of the event text; therefore, the accuracy of event category identification can be effectively improved.

In the foregoing embodiment, an event category determining method is provided, and correspondingly, the present application further provides an event category determining apparatus. The apparatus corresponds to an embodiment of the method described above.

Second embodiment

Please refer to fig. 7, which is a schematic diagram of an embodiment of an event category determining apparatus of the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

The present application further provides an event category determination device, including:

an element type determining unit 701, configured to determine words included in the event text to be classified and event element types of the words;

an event feature extraction unit 702, configured to extract, from the event feature extraction sub-network included in the event classification model, an event feature including text semantic information and event element type information according to a word vector of a word and an element vector of the event element type;

an event classification determining unit 703 is configured to determine an event classification of the event text according to the event feature by using an event classification prediction sub-network included in the event classification model.

Optionally, the element type determining unit 701 is specifically configured to, for each word, use the word as a target word, and determine a text segment including the target word and adjacent words thereof; acquiring word vectors of a target word and adjacent words; determining word position vectors of the target word and the adjacent words according to word position information of the adjacent words in the text segment; and aiming at each target word, determining the event element type of the target word according to the word vector of the target word and adjacent words thereof and the word position vector by an event element type prediction model.

Optionally, the event feature extraction unit 702 is specifically configured to determine, according to a word vector of a word and the element vector of the event element type, a word vector of the word that includes element type information; and extracting the event features according to the word vector comprising the element type information through the event feature extraction sub-network.

Optionally, the event element types include: time, place, person, event.

Optionally, the event categories include: location type, time period, gender of the person, event type.

Third embodiment

Please refer to fig. 8, which is a diagram illustrating an embodiment of an electronic device according to the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

An electronic device of the present embodiment includes: a processor 801 and a memory 802; the memory is used for storing a program for realizing the event type determination method, and after the device is powered on and the program for realizing the event confirmation method is run by the processor, the following steps are executed: determining words included in the event text to be classified and event element types of the words; extracting a sub-network through event features included by an event classification model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element vector of the event element type; and determining the event category of the event text according to the event characteristics through an event category prediction sub-network included in the event classification model.

In the first embodiment described above, an event type determining method is provided, and correspondingly, the present application also provides an equipment failure occurrence mode determining method.

Fourth embodiment

Please refer to fig. 9, which is a flowchart illustrating an embodiment of a method for determining an equipment failure mode according to the present application, wherein an execution main body of the method includes an equipment failure mode determining apparatus. Since the method embodiment corresponds to the method embodiment of the first embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the first embodiment. The method embodiments described below are merely illustrative.

The method for determining the equipment failure mode comprises the following steps:

step S1201: and acquiring a text related to the equipment failure.

For example, the equipment failure report includes information of pre-accident conditions, accident passing and handling conditions, accident reasons, accident loss conditions and the like of certain equipment (such as a certain model of valve-regulated sealed lead-acid battery pack) of a certain enterprise (such as a power enterprise).

Step S1203: determining words included in the text and equipment fault element types of the words.

The type of the fault element of the equipment can be climate, time, place, fault phenomenon and the like.

In one example, the device fault element type may be determined using the following steps: 1) aiming at each word, taking the word as a target word, and determining a text segment comprising the target word and adjacent words thereof; 2) acquiring word vectors of a target word and adjacent words; determining word position vectors of the target word and the adjacent words according to word position information of the adjacent words in the text segment; 3) and aiming at each target word, determining the equipment fault element type of the target word according to the word vector of the target word and adjacent words thereof and the word position vector by an equipment fault element type prediction model.

In this case, the method may further include the steps of: and learning to obtain the equipment fault element type prediction model from the corresponding relation between the equipment fault text for training and the equipment fault element type marking information of the words included in the equipment fault text for training.

Step S1205: and extracting the equipment fault features comprising text semantic information and equipment fault element type information according to the word vector of the words and the element type vector of the equipment fault element types through the equipment fault feature extraction sub-network included in the fault element determination model.

In one example, step S1205 may include the following sub-steps: 1) determining word vectors of words including equipment fault element type information according to the word vectors of the words and the element type vectors; 2) and extracting the equipment fault characteristics through the equipment fault characteristic extraction sub-network according to the word vector comprising the equipment fault element type information.

In this case, the method may further include the steps of: and learning to obtain the fault element determination model from the corresponding relation between the training equipment fault text and the equipment fault element marking information.

Step S1207: and determining equipment fault element information according to the equipment fault characteristics through the fault element prediction sub-network included in the fault element determination model.

The equipment failure factor information may include at least one of a climate type, a location type, a time period, a failure type, and the like. Specifically, the climate type may be drought, rainstorm, or cold; the location types include: north, south; the time period includes: morning, noon, evening; the fault types include: leakage and large power consumption.

Step S1209: and determining equipment fault occurrence mode information according to the equipment fault element information.

After determining the element information of the equipment failure event, the equipment failure occurrence mode information can be determined according to the information. For example, the multiple modes of a certain device in a power enterprise are: large power consumption in dry weather, and the like. For another example, the device is a server, and the multiple mode of server failure is as follows: an access volume exceeding x times is prone to down, etc.

In one example, the method may further comprise the steps of: and determining relevant parameters of the server operation according to the equipment failure occurrence mode information. For example, the access amount of the server is limited to a maximum of y times, and so on.

As can be seen from the foregoing embodiments, the method for determining an equipment failure occurrence mode provided in the embodiments of the present application obtains an equipment failure related text; determining words included in the text and equipment fault element types of the words; extracting a sub-network through the equipment fault feature included in the fault element determination model, and extracting equipment fault features including text semantic information and equipment fault element type information according to a word vector of a word and an element type vector of the equipment fault element type; determining equipment fault element information according to the equipment fault characteristics through a fault element prediction sub-network included in a fault element determination model; determining equipment fault occurrence mode information according to the equipment fault element information; the processing mode can automatically determine the equipment failure occurrence mode, help to find the situation with multiple equipment failures and take preventive measures; therefore, the determination efficiency of the equipment failure occurrence mode can be effectively improved.

In the foregoing embodiment, a method for determining an equipment failure occurrence mode is provided, and correspondingly, an apparatus failure occurrence mode determining device is also provided in the present application. The apparatus corresponds to an embodiment of the method described above.

Fifth embodiment

Please refer to fig. 10, which is a schematic diagram of an embodiment of the device failure occurrence mode determining apparatus of the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

The present application further provides an apparatus for determining an occurrence mode of an equipment failure, including:

a trouble text acquisition unit 1001 configured to acquire an apparatus trouble related text;

an element type determining unit 1002, configured to determine a word included in the text and an equipment failure element type of the word;

a fault feature extraction unit 1003, configured to extract a sub-network from the device fault features included in the fault element determination model, and extract device fault features including text semantic information and device fault element type information according to the word vector of the word and the element type vector of the device fault element type;

a failure element determination unit 1004 for determining device failure element information from the device failure characteristics by using a failure element prediction sub-network included in a failure element determination model;

a failure mode determination unit 1005 for determining the device failure occurrence mode information based on the device failure element information.

Optionally, the device failure element types include: weather, time, location, failure.

Optionally, the device failure element information includes: climate type, location type, time period, fault type.

Optionally, the climate types include: drought, rainstorm, cold; the location types include: north, south; the time period includes: morning, noon, evening; the fault types include: leakage and large power consumption.

Optionally, the element type determining unit 1002 is specifically configured to, for each word, use the word as a target word, and determine a text segment including the target word and adjacent words thereof; acquiring word vectors of a target word and adjacent words; determining word position vectors of the target word and the adjacent words according to word position information of the adjacent words in the text segment; and aiming at each target word, determining the equipment fault element type of the target word according to the word vector of the target word and adjacent words thereof and the word position vector by an equipment fault element type prediction model.

Optionally, the apparatus further comprises:

and the first model construction unit is used for learning and obtaining the equipment fault element type prediction model from the corresponding relation between the equipment fault text for training and the equipment fault element type marking information of the words included in the equipment fault text for training.

Optionally, the fault feature extraction unit 1003 is specifically configured to determine, according to a word vector of a word and the element type vector, a word vector of the word, which includes device fault element type information; and extracting the equipment fault characteristics through the equipment fault characteristic extraction sub-network according to the word vector comprising the equipment fault element type information.

Optionally, the apparatus further comprises:

and the second model construction unit is used for learning and obtaining the fault element determination model from the corresponding relation between the training equipment fault text, the training equipment fault text packet and the equipment fault element marking information.

Optionally, the apparatus further comprises:

and the equipment operation parameter determining unit is used for determining relevant parameters of equipment operation according to the equipment fault occurrence mode information.

Optionally, the device includes a server or a power device.

Sixth embodiment

Please refer to fig. 11, which is a diagram illustrating an embodiment of an electronic device according to the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

An electronic device of the present embodiment includes: a processor 1101 and a memory 1102; the memory is used for storing a program for realizing the device failure occurrence mode determination method, and after the device is powered on and runs the program of the method through the processor, the following steps are executed: acquiring a text related to equipment failure; determining words included in the text and equipment fault element types of the words; extracting a sub-network through the equipment fault feature included in the fault element determination model, and extracting equipment fault features including text semantic information and equipment fault element type information according to a word vector of a word and an element type vector of the equipment fault element type; determining equipment fault element information according to the equipment fault characteristics through a fault element prediction sub-network included in a fault element determination model; and determining equipment fault occurrence mode information according to the equipment fault element information.

In the first embodiment, an event type determining method is provided, and correspondingly, an event occurrence mode determining method is also provided in the present application.

Seventh embodiment

Please refer to fig. 12, which is a flowchart illustrating an embodiment of a method for determining an event occurrence mode according to the present application, wherein an executing entity of the method includes an event occurrence mode determining apparatus. Since the method embodiment corresponds to the method embodiment of the first embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the first embodiment. The method embodiments described below are merely illustrative.

The event occurrence mode determining method provided by the application comprises the following steps:

step S1201: and acquiring event related text.

Step S1203: determining words included in the text and event element types of the words.

The event element types include, but are not limited to: time, place, victim, suspect, action, and the like.

Step S1205: and extracting the event features comprising text semantic information and event element type information according to the word vector of the words and the element type vector of the event element types through the event feature extraction sub-network included in the event element determination model.

Step S1207: and determining event element information according to the event characteristics by using the event element prediction sub-network included in the event element determination model.

Step S1209: and determining event occurrence mode information according to the event element information.

The event element information includes, but is not limited to: type of location of case, time period of case, sex of victim, type of event, etc. The location type may be a public transportation facility, a restaurant, a school, a home, etc.; the time period can be morning, noon and evening; the victim, which may be male or female; the type of event may be a theft, personal injury, fraud, or the like.

In one example, the event occurrence pattern information includes: the mode of occurrence of events related to public transport facilities, such as the case of theft easily occurs during the rush hour when buses go on and off the work; accordingly, the method may further comprise the steps of: and determining the dispatching information of the public transport personnel according to the event occurrence mode information. For example, bus security personnel are added during rush hours.

As can be seen from the foregoing embodiments, the event occurrence mode determining method provided in the embodiments of the present application obtains an event-related text; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information; the processing mode can automatically extract elements from the event text, classify the mode condition of the case, automatically determine the event occurrence mode, help to determine the frequent situation of the event and take preventive measures; therefore, the determination efficiency of the event occurrence mode can be effectively improved.

In the foregoing embodiment, an event occurrence mode determining method is provided, and correspondingly, the present application also provides an event occurrence mode determining apparatus. The apparatus corresponds to an embodiment of the method described above.

Eighth embodiment

Please refer to fig. 13, which is a schematic diagram of an embodiment of an event occurrence mode determining apparatus of the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

The present application further provides an event occurrence pattern determination apparatus, including:

an event text acquiring unit 1301, configured to acquire an event related text;

an element type determining unit 1302, configured to determine a word included in the text and an event element type of the word;

an event feature extraction unit 1303, configured to extract, from the event feature extraction sub-network included in the event element determination model, an event feature including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type;

an event element determination unit 1304 for determining event element information from the event feature by using an event element prediction sub-network included in the event element determination model;

an event occurrence mode determining unit 1305, configured to determine event occurrence mode information according to the event element information.

Optionally, the event element types include: time, place, victim, event.

Optionally, the event element information includes: location type, time period, victim gender, event type.

Optionally, the location types include: public transportation facilities, restaurants, homes; the time period includes: morning, noon, evening; the victim includes: male, female; the event types include: theft, personal injury, fraud.

Optionally, the event occurrence mode information includes: patterns of event occurrences related to public transportation facilities; correspondingly, the device can also comprise the following units:

and the scheduling information determining unit is used for determining the bus personnel scheduling information according to the event occurrence mode information.

Ninth embodiment

Please refer to fig. 14, which is a diagram illustrating an embodiment of an electronic device according to the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

An electronic device of the present embodiment includes: a processor 1401 and a memory 1402; the memory is used for storing a program for realizing the event occurrence mode determining method, and after the device is powered on and runs the program for realizing the event occurrence mode determining method through the processor, the following steps are executed: acquiring an event related text; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information according to the event characteristics through an event element prediction sub-network included in an event element determination model; and determining event occurrence mode information according to the event element information.

In the first embodiment described above, an event type determining method is provided, and correspondingly, the application further provides an enterprise acquisition event occurrence pattern determining method.

Tenth embodiment

Please refer to fig. 15, which is a flowchart illustrating an embodiment of a method for determining an occurrence pattern of an enterprise acquisition event according to the present application, wherein an executing entity of the method includes an enterprise acquisition event occurrence pattern determining apparatus. Since the method embodiment corresponds to the method embodiment of the first embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the first embodiment. The method embodiments described below are merely illustrative.

The method for determining the occurrence mode of the enterprise acquisition event comprises the following steps:

step S1501: acquiring a text related to an enterprise purchase event;

step S1503: determining words included in the text and the type of the enterprise purchase event elements of the words;

the enterprise acquisition event element type can be acquisition amount, acquisition reason, accident reason of the supplier, accident time of the supplier and the like.

Step S1505: extracting a sub-network of enterprise acquisition event features included by an enterprise acquisition event element determination model, and extracting enterprise acquisition event features including text semantic information and enterprise acquisition event element type information according to a word vector of a word and an element type vector of an enterprise acquisition event element type;

step S1507: determining enterprise acquisition event element information according to the characteristics of the enterprise acquisition event by an enterprise acquisition event element prediction sub-network included in an enterprise acquisition event element determination model;

step S1509: and determining the occurrence mode information of the enterprise acquisition event according to the element information of the enterprise acquisition event.

The enterprise procurement event element information includes, but is not limited to: the purchase amount level, the purchase reason type, the accident time type and the like.

For example, the elements of the company-associated purchase event concerned by the similar company such as the automobile, etc. are extracted, the purchase event relates to the elements such as the amount of money, the reason of purchase, the time of the accident of the supplier, the reason, etc., and the classification of the purchase event is performed, for example, the classification is performed according to the size of the amount of money or the reason, etc.

As can be seen from the foregoing embodiments, the method for determining an equipment failure occurrence mode provided in the embodiments of the present application obtains a text related to an enterprise acquisition event; determining words included in the text and the type of the enterprise purchase event elements of the words; extracting a sub-network of enterprise acquisition event features included by an enterprise acquisition event element determination model, and extracting enterprise acquisition event features including text semantic information and enterprise acquisition event element type information according to a word vector of a word and an element type vector of an enterprise acquisition event element type; determining enterprise acquisition event element information according to the characteristics of the enterprise acquisition event by an enterprise acquisition event element prediction sub-network included in an enterprise acquisition event element determination model; determining the occurrence mode information of the enterprise acquisition event according to the element information of the enterprise acquisition event; the processing mode enables the elements of the enterprise co-purchasing event to be automatically extracted, classifies the mode condition of the co-purchasing event, automatically determines the occurrence mode of the co-purchasing event and helps to determine the situation of multiple co-purchasing events so as to specify the enterprise co-purchasing plan; therefore, the determining efficiency of the enterprise co-purchase event occurrence mode can be effectively improved.

In the foregoing embodiment, a method for determining an occurrence mode of an enterprise acquisition event is provided, and correspondingly, an apparatus for determining an occurrence mode of an enterprise acquisition event is also provided in the present application. The apparatus corresponds to an embodiment of the method described above.

Eleventh embodiment

Please refer to fig. 16, which is a schematic diagram of an embodiment of an enterprise acquisition event occurrence pattern determining apparatus according to the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

The present application additionally provides an enterprise acquisition event occurrence pattern determining apparatus, comprising:

an event text acquiring unit 1601, configured to acquire an enterprise acquisition event related text;

an element type determining unit 1602, configured to determine a word included in the text and an enterprise acquisition event element type of the word;

an event feature extracting unit 1603, configured to extract, according to the word vector of the word and the element type vector of the enterprise acquisition event element type, an enterprise acquisition event feature including text semantic information and enterprise acquisition event element type information through an enterprise acquisition event feature extraction sub-network included in the enterprise acquisition event element determination model;

an event element determining unit 1604, configured to determine, according to the characteristics of the enterprise acquisition event, enterprise acquisition event element information by using an enterprise acquisition event element prediction subnetwork included in the enterprise acquisition event element determination model;

an event occurrence mode determining unit 1605, configured to determine the enterprise acquisition event occurrence mode information according to the enterprise acquisition event element information.

Optionally, the enterprise acquisition event element types include: the purchase amount, the purchase reason, the accident reason of the supplier and the accident time of the supplier.

Optionally, the enterprise acquisition event element information includes: the purchase amount grade, the purchase reason type, the accident reason type and the accident time type.

Twelfth embodiment

Please refer to fig. 17, which is a diagram illustrating an embodiment of an electronic device according to the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

An electronic device of the present embodiment includes: a processor 1701 and a memory 1702; the memory is used for storing a program for realizing the method for determining the occurrence mode of the enterprise acquisition event, and after the equipment is powered on and runs the program for determining the occurrence mode of the enterprise acquisition event through the processor, the following steps are executed: acquiring a text related to an enterprise purchase event; determining words included in the text and the type of the enterprise purchase event elements of the words; extracting a sub-network of enterprise acquisition event features included by an enterprise acquisition event element determination model, and extracting enterprise acquisition event features including text semantic information and enterprise acquisition event element type information according to a word vector of a word and an element type vector of an enterprise acquisition event element type; determining enterprise acquisition event element information according to the characteristics of the enterprise acquisition event by an enterprise acquisition event element prediction sub-network included in an enterprise acquisition event element determination model; and determining the occurrence mode information of the enterprise acquisition event according to the element information of the enterprise acquisition event.

In the first embodiment, an event type determining method is provided, and correspondingly, a method for constructing an event element type prediction model is also provided. The method corresponds to the embodiment of the method described above.

Thirteenth embodiment

Please refer to fig. 18, which is a flowchart illustrating an embodiment of a method for constructing an event element type prediction model according to the present application, wherein an execution subject of the method includes an event element type prediction model constructing apparatus. Since the method embodiment is a part of the method embodiment of the first embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the method embodiment. The method embodiments described below are merely illustrative.

The method for constructing the event element type prediction model comprises the following steps:

step S1801: a training sample set is obtained.

The training samples comprise training event texts, words included in the training event texts and corresponding relations among event element type marking information of the words.

Step S1803: and constructing a network structure of the model.

Wherein the network structure comprises a feature extraction sub-network and an event element type prediction sub-network; the feature extraction sub-network is used for extracting event features used for predicting event element types according to word vectors of words of the event texts; and the event element type prediction sub-network is used for acquiring the predicted value of the event element type of the word according to the event characteristics.

Step S1805: and training the network parameters of the model according to the training sample set to obtain an event element type prediction model.

In one example, the feature extraction sub-network is specifically configured to extract the event features according to word vectors of words of the event text and their neighboring words and word position vectors, where the event features include context semantic information of the words.

As can be seen from the above embodiments, the event element type prediction model construction method provided in the embodiments of the present application obtains a training sample set; the training samples comprise training event texts, words included in the training event texts and corresponding relations among event element type marking information of the words; constructing a network structure of the model; wherein the network structure comprises a feature extraction sub-network and an event element type prediction sub-network; the feature extraction sub-network is used for extracting event features used for predicting event element types according to word vectors of words of the event texts; the event element type prediction sub-network is used for acquiring a predicted value of an event element type of a word according to the event characteristics; training the network parameters of the model according to the training sample set to obtain an event element type prediction model; the processing mode can combine the word body semantic information of the event text words and the information of the peripheral text to construct an event element type prediction model; therefore, the prediction accuracy of the event element type prediction model can be effectively improved.

In the foregoing embodiment, a method for constructing an event element type prediction model is provided, and correspondingly, an apparatus for constructing an event element type prediction model is also provided in the present application. The apparatus corresponds to an embodiment of the method described above.

Fourteenth embodiment

Please refer to fig. 19, which is a schematic diagram of an embodiment of an event element type prediction model construction apparatus according to the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

The present application further provides an event element type prediction model construction device, including:

a data obtaining unit 1901, configured to obtain a training sample set; the training samples comprise training event texts, words included in the training event texts and corresponding relations among event element type marking information of the words;

a network construction unit 1902, configured to construct a network structure of the model; wherein the network structure comprises a feature extraction sub-network and an event element type prediction sub-network; the feature extraction sub-network is used for extracting event features used for predicting event element types according to word vectors of words of the event texts; the event element type prediction sub-network is used for acquiring a predicted value of an event element type of a word according to the event characteristics;

and a network training unit 1903, configured to train the network parameters of the model according to the training sample set, so as to obtain an event element type prediction model.

Fifteenth embodiment

Please refer to fig. 20, which is a diagram illustrating an embodiment of an electronic device according to the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

An electronic device of the present embodiment includes: a processor 2001 and a memory 2002; the memory is used for storing a program for realizing the event type prediction model building method, and after the device is powered on and runs the program for realizing the event type prediction model building method through the processor, the following steps are executed: acquiring a training sample set; the training sample comprises a training event text, words included in the training event text and corresponding relations among event element type marking information of the words; constructing a network structure of the model; wherein the network structure comprises a feature extraction sub-network and an event element type prediction sub-network; the feature extraction sub-network is used for extracting event features used for predicting event element types according to word vectors of words of the event texts; the event element type prediction sub-network is used for acquiring a predicted value of an event element type of a word according to the event characteristics; and training the network parameters of the model according to the training sample set to obtain an event element type prediction model.

In the first embodiment, an event classification determining method is provided, and correspondingly, an event classification model constructing method is also provided in the present application. The method corresponds to the embodiment of the method described above.

Sixteenth embodiment

Please refer to fig. 21, which is a flowchart illustrating an embodiment of a method for constructing an event classification model according to the present application, wherein an executing entity of the method includes an event classification model constructing apparatus. Since the method embodiment is a part of the method embodiment of the first embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the method embodiment. The method embodiments described below are merely illustrative.

The method for constructing the event classification model comprises the following steps:

step S2101: a training sample set is obtained.

The training sample comprises corresponding relations between training event texts and event category marking information.

In one example, the training sample includes event element type tagging information that may also be event text words. By adopting the method, the model accuracy can be effectively improved.

Step S2103: constructing a network structure of the model; wherein the network structure comprises an event feature extraction sub-network and an event category prediction sub-network; the feature extraction sub-network is used for extracting event features according to the word vectors of all words of the event text and the element vectors of the event element types with the words as target words, wherein the features comprise text semantic information and event element type information; and the event type prediction sub-network is used for acquiring the predicted value of the event type of the event text according to the event characteristics.

Step S2104: and training the network parameters of the model according to the training sample set to obtain an event classification model.

As can be seen from the above embodiments, the event classification model construction method provided in the embodiments of the present application obtains a training sample set; the training sample comprises corresponding relations between training event texts and event category marking information; constructing a network structure of the model; wherein the network structure comprises an event feature extraction sub-network and an event category prediction sub-network; the feature extraction sub-network is used for extracting event features according to the word vectors of all words of the event text and the element vectors of the event element types with the words as target words, wherein the features comprise text semantic information and event element type information; the event type prediction sub-network is used for acquiring a predicted value of an event type of an event text according to the event characteristics; training the network parameters of the model according to the training sample set to obtain an event classification model; the processing mode can combine the word body semantic information and the event element type information of the event text words to construct an event classification prediction model; therefore, the prediction accuracy of the event classification model can be effectively improved.

In the foregoing embodiment, an event classification model construction method is provided, and correspondingly, the present application also provides an event classification model construction device. The apparatus corresponds to an embodiment of the method described above.

Seventeenth embodiment

Please refer to fig. 22, which is a schematic diagram of an embodiment of an event classification model construction apparatus according to the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

a data acquisition unit 2201 for acquiring a training sample set; the training sample comprises corresponding relations between training event texts and event category marking information;

a network construction unit 2202 for constructing a network structure of the model; wherein the network structure comprises an event feature extraction sub-network and an event category prediction sub-network; the feature extraction sub-network is used for extracting event features according to the word vectors of all words of the event text and the element vectors of the event element types with the words as target words, wherein the features comprise text semantic information and event element type information; the event type prediction sub-network is used for acquiring a predicted value of an event type of an event text according to the event characteristics;

and the network training unit 2203 is configured to train the network parameters of the model according to the training sample set to obtain an event classification model.

Eighteenth embodiment

Please refer to fig. 23, which is a diagram illustrating an embodiment of an electronic device according to the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

An electronic device of the present embodiment includes: a processor 2301 and memory 2302; the memory is used for storing a program for realizing the event classification model building method, and after the device is powered on and runs the program for realizing the event classification model building method through the processor, the following steps are executed: acquiring a training sample set; the training sample comprises corresponding relations between training event texts and event category marking information; constructing a network structure of the model; wherein the network structure comprises an event feature extraction sub-network and an event category prediction sub-network; the feature extraction sub-network is used for extracting event features according to the word vectors of all words of the event text and the element vectors of the event element types with the words as target words, wherein the features comprise text semantic information and event element type information; the event type prediction sub-network is used for acquiring a predicted value of an event type of an event text according to the event characteristics; and training the network parameters of the model according to the training sample set to obtain an event classification model.

In the first embodiment, an event type determining method is provided, and correspondingly, a traffic accident occurrence mode determining method is also provided.

Nineteenth embodiment

Please refer to fig. 24, which is a flowchart illustrating an embodiment of a traffic accident occurrence mode determining method according to the present application, wherein an executing body of the method includes a traffic accident occurrence mode determining device. Since the method embodiment corresponds to the method embodiment of the first embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the first embodiment. The method embodiments described below are merely illustrative.

The traffic accident occurrence mode determining method provided by the application comprises the following steps:

step S2401: acquiring a text related to a traffic accident;

step S2403: determining words included in the text and traffic accident element types of the words;

the traffic accident element types can be accident occurrence time, accident occurrence place, climate when the accident occurs, accident occurrence mode and the like.

Step S2405: extracting traffic accident features including text semantic information and traffic accident element type information according to a word vector of a word and an element type vector of a traffic accident element type through a traffic accident feature extraction sub-network included in a traffic accident element determination model;

step S2407: determining traffic accident factor information according to the traffic accident characteristics through a traffic accident factor prediction sub-network included in a traffic accident factor determination model;

step S2409: and determining traffic accident occurrence mode information according to the traffic accident factor information.

The traffic accident element information includes, but is not limited to: the accident occurrence time period, the accident occurrence place type, the accident occurrence climate type, the accident occurrence mode type and the like. The accident site types can be sharp turning on mountain roads, ascending/descending slopes, places near schools, places near business markets, intersections without traffic lights and the like; the time period can be morning, commute/off-duty time period, noon and evening; the accident occurrence mode types can be rear-end accidents, straight-ahead accidents, overtaking accidents, left/right turning accidents, curve accidents, ramp accidents, narrow-lane accidents, meeting accidents, overtaking accidents, parking accidents, pedestrian rubbing and the like.

For example, the traffic accident occurrence pattern information may be: the intersection without traffic lights is easy to have direct accidents at the peak time of going to work and can be: the intersection at the position of the slope ascending first and the slope descending later is easy to have rear-end collision accidents caused by untimely braking in rainy and foggy weather.

In one example, the factor information includes: traffic accident grade information; the accident grade can be a light accident, a general accident and a major accident. Accordingly, the method may further comprise the steps of: determining traffic accident grade information of a plurality of traffic accident related texts; and determining the quantity of the traffic accidents of each traffic accident grade according to the traffic accident grade information.

As can be seen from the above embodiments, the traffic accident occurrence mode determination method provided by the embodiment of the present application obtains a text related to a traffic accident; determining words included in the text and traffic accident element types of the words; extracting traffic accident features including text semantic information and traffic accident element type information according to a word vector of a word and an element type vector of a traffic accident element type through a traffic accident feature extraction sub-network included in a traffic accident element determination model; determining traffic accident factor information according to the traffic accident characteristics through a traffic accident factor prediction sub-network included in a traffic accident factor determination model; determining traffic accident occurrence mode information according to the traffic accident element information; the processing mode can automatically extract elements of the traffic accident case of the traffic management department, classify the mode situation of the accident, automatically determine the traffic accident occurrence mode, help to determine the situation of frequent traffic accidents, and take preventive measures; therefore, the determining efficiency and the determining accuracy of the traffic accident occurrence mode can be effectively improved.

In the above embodiment, a traffic accident occurrence mode determining method is provided, and correspondingly, the application also provides a traffic accident occurrence mode determining device. The apparatus corresponds to an embodiment of the method described above.

Twentieth embodiment

Please refer to fig. 25, which is a schematic diagram of an embodiment of a traffic accident occurrence mode determination apparatus of the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

The present application additionally provides a traffic accident occurrence pattern determination apparatus, including:

an accident file acquiring unit 2501, configured to acquire a text related to a traffic accident;

an element type determining unit 2502, configured to determine words included in the text and traffic accident element types of the words;

an accident feature determination unit 2503, configured to extract traffic accident features including text semantic information and traffic accident element type information according to the word vector of the word and the element type vector of the traffic accident element type through a traffic accident feature extraction sub-network included in the traffic accident element determination model;

an accident element determination unit 2504 for determining traffic accident element information based on the traffic accident characteristics by using the traffic accident element prediction sub-network included in the traffic accident element determination model;

an accident mode determining unit 2506, configured to determine traffic accident occurrence mode information according to the traffic accident factor information.

Optionally, the element information includes: traffic accident grade information; correspondingly, the device can also comprise the following units:

and the traffic accident quantity determining unit is used for determining the traffic accident quantity of each traffic accident grade according to the traffic accident grade information.

Twenty-first embodiment

Please refer to fig. 26, which is a diagram illustrating an embodiment of an electronic device according to the present application. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

An electronic device of the present embodiment includes: a processor 2501 and a memory 2502; the memory is used for storing a program for realizing the traffic accident occurrence mode determining method, and after the equipment is powered on and runs the program for the traffic accident occurrence mode determining method through the processor, the following steps are executed: acquiring a text related to a traffic accident; determining words included in the text and traffic accident element types of the words; extracting traffic accident features including text semantic information and traffic accident element type information according to a word vector of a word and an element type vector of a traffic accident element type through a traffic accident feature extraction sub-network included in a traffic accident element determination model; determining traffic accident factor information according to the traffic accident characteristics through a traffic accident factor prediction sub-network included in a traffic accident factor determination model; and determining traffic accident occurrence mode information according to the traffic accident factor information.

In the nineteenth embodiment, a traffic accident occurrence mode determining method is provided, and correspondingly, the application further provides a traffic information broadcasting system.

Twenty-second embodiment

Please refer to fig. 27, which is a schematic diagram of an embodiment of a traffic information broadcasting system according to the present application. Since the system embodiment corresponds to the method embodiment of nineteen embodiments, the description is relatively simple, and the relevant points can be referred to the partial description of nineteen embodiments. The system embodiments described below are merely illustrative.

The application provides a traffic information reports system includes: a first server 2701, a second server 2702, and a client 2703.

Please refer to fig. 28, which is a schematic diagram of device interaction in an embodiment of a traffic information broadcast system provided in the present application. The first server is used for acquiring a text related to the traffic accident; determining words included in the text and traffic accident element types of the words; extracting traffic accident features including text semantic information and traffic accident element type information according to a word vector of a word and an element type vector of a traffic accident element type through a traffic accident feature extraction sub-network included in a traffic accident element determination model; determining traffic accident factor information according to the traffic accident characteristics through a traffic accident factor prediction sub-network included in a traffic accident factor determination model; and determining traffic accident occurrence mode information according to the traffic accident factor information. The first service end can send the determined traffic accident occurrence mode information to the second service end.

The second server can be deployed in a map server (such as a Gauss map) and used for collecting various traffic accident occurrence mode information and determining accident multi-occurrence section information and accident occurrence condition information according to the traffic accident occurrence mode information. In addition, accident type information can be determined according to the traffic accident occurrence mode information. Table 1 shows the correspondence among the accident-prone-segment information, the accident occurrence condition information, and the accident type information of the present embodiment.

TABLE 1 set of correspondences

As can be seen from table 1, if the first server determines that the traffic accident occurrence mode information is: the intersection without the traffic lights is easy to have a straight-ahead accident in the peak time of going to work and off duty, and the second server can determine the accident multi-occurrence section according to the map data, and the method comprises the following steps: intersection 023, intersection 452, and the like, which are not installed with traffic lights; the accident occurrence conditions include: the peak time of going to work and off work; the types of traffic accidents are: and (5) a straight-going accident. In addition, the second server can determine that the sections can also correspond to other accident occurrence conditions and traffic accident types according to other mode information, wherein the accident occurrence conditions are as follows: at night, the traffic accident types are: a passing accident, etc.

For another example, the first service end determines that the traffic accident occurrence mode information is: the following steps are that the following accident caused by untimely braking is easy to occur at the intersection at the position of the downhill after the uphill in rainy and foggy weather (or at peak time of going to work and going to work), and the second server can determine the accident multi-occurrence section according to the map data, and the following steps are included: intersections 045, 231 and the like go up the slope first and then go down the slope; the accident occurrence conditions include: rain and fog weather (or rush hour to work); the types of traffic accidents are: rear-end accidents, etc.

For another example, the first service end determines that the traffic accident occurrence mode information is: the red light running accident caused by the fact that the signal light cannot be seen clearly occurs easily in sunny days in summer and between 8 and 18 points when the road section is not shielded by the building, and then the second server side can determine the accident multi-occurrence section according to the map data and comprises the following steps: road segment 002, road segment 565, and so on; the accident occurrence conditions include: in sunny days in summer, 8-18 points; the types of traffic accidents are: red light running caused by unclear signal light, and the like.

The second server is also used for determining the current driving position of the vehicle, and if the distance between the position of the vehicle and the accident multi-occurrence section is smaller than a distance threshold value and the accident occurrence condition is satisfied, determining traffic accident early warning information and sending the early warning information to the client; correspondingly, the client can be used for displaying the early warning information to remind a driver of paying attention to traffic safety, or the unmanned vehicle can automatically reduce the speed according to the early warning information, and the like.

The distance threshold may be determined according to the service requirement or experience, such as set to 500 meters, etc.

The client can be deployed in the automatic driving system of the vehicle and can also be deployed in the mobile communication equipment.

In one example, the client is a mobile APP deployed on a mobile communication device, such as a gold map APP, which is opened by a driver during driving, and through which the current driving position of the vehicle can be determined; the second server side obtains the position information, judges whether the distance between the current driving position of the vehicle and the accident frequently-occurring section is smaller than a distance threshold value according to the position information, and determines an accident occurrence condition according to traffic accident occurrence mode information related to the accident frequently-occurring section, wherein the accident occurrence condition is rain and fog weather; correspondingly, the second server side obtains the current climate information of the section with the multiple accidents, judges whether the accident occurrence condition is satisfied or not according to the climate information, determines the traffic accident early warning information if the accident occurrence condition is satisfied, and sends the early warning information to the client side if the traffic accident early warning information is satisfied, such as that a straight-ahead accident (or frequent accident) is easy to occur at an intersection 500 meters ahead in rainy days, and the client side is requested to slow down and walk. In another example, the client is an automatic driving system in an automatic driving vehicle, and the system is used for receiving the early warning information and automatically adjusting the driving mode of the vehicle according to the early warning information.

As can be seen from the above embodiments, the traffic information broadcasting system provided by the embodiment of the application obtains the text related to the traffic accident through the first server; determining words included in the text and traffic accident element types of the words; extracting traffic accident features including text semantic information and traffic accident element type information according to a word vector of a word and an element type vector of a traffic accident element type through a traffic accident feature extraction sub-network included in a traffic accident element determination model; determining traffic accident factor information according to the traffic accident characteristics through a traffic accident factor prediction sub-network included in a traffic accident factor determination model; determining traffic accident occurrence mode information according to the traffic accident element information; determining accident multi-occurrence section information and accident occurrence condition information according to the traffic accident occurrence mode information through a second server; determining the position information of a vehicle to be early-warned; if the distance between the vehicle position and the accident multi-occurrence section is smaller than the distance threshold value and the accident occurrence condition is established, determining traffic accident early warning information, and sending the early warning information to a client associated with the vehicle to be early warned; the client receives the early warning information so as to adjust the driving mode of the vehicle according to the early warning information; the processing mode can remind the driver of paying attention to safe driving in time; therefore, driving safety can be effectively improved.

Twenty-third embodiment

The application also provides a traffic information broadcasting method, and an execution main body of the method comprises a traffic information broadcasting device. Since this method embodiment corresponds to the method embodiment of twenty-two, the description is relatively simple, and reference may be made to the partial description of twenty-two. The method embodiments described below are merely illustrative.

The traffic information broadcasting method provided by the application comprises the following steps:

step 1: and determining the information of the accident multi-occurrence section and the accident occurrence condition information according to the traffic accident occurrence mode information.

Step 2: and determining the position information of the vehicle to be early-warned.

And step 3: and if the distance between the vehicle position and the accident multi-occurrence section is smaller than the distance threshold value and the accident occurrence condition is satisfied, determining traffic accident early warning information and sending the early warning information to a client associated with the vehicle to be early warned. As can be seen from the above embodiments, the traffic information broadcasting method provided by the embodiment of the present application determines the accident multi-occurrence section information and the accident occurrence condition information according to the traffic accident occurrence mode information; determining the position information of a vehicle to be early-warned; if the distance between the vehicle position and the accident multi-occurrence section is smaller than the distance threshold value and the accident occurrence condition is established, determining traffic accident early warning information; sending the early warning information to a client side associated with the vehicle to be early warned so as to adjust the driving mode of the vehicle according to the early warning information; the processing mode can remind the driver of paying attention to safe driving in time; therefore, driving safety can be effectively improved.

In the above embodiment, a traffic information broadcasting method is provided, and correspondingly, the application also provides a traffic information broadcasting device. The apparatus corresponds to an embodiment of the method described above.

Twenty-fourth embodiment

The application also provides a traffic information broadcasting device. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

The application additionally provides a traffic information reports device, includes:

Twenty-fifth embodiment

The application also provides an electronic device. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

An electronic device of the present embodiment includes: a processor and a memory; the memory is used for storing a program for realizing the traffic information broadcasting method, and after the equipment is powered on and runs the program for realizing the traffic information broadcasting method through the processor, the following steps are executed: determining accident multi-occurrence section information and accident occurrence condition information according to the traffic accident occurrence mode information; determining the position information of a vehicle to be early-warned; if the distance between the vehicle position and the accident multi-occurrence section is smaller than the distance threshold value and the accident occurrence condition is established, determining traffic accident early warning information; and sending the early warning information to a client associated with the vehicle to be early warned.

Twenty-sixth embodiment

The application also provides an embodiment of the traffic information broadcasting method, and an execution main body of the method comprises a traffic information broadcasting device. Since this method embodiment corresponds to the method embodiment of twenty-two, the description is relatively simple, and reference may be made to the partial description of twenty-two. The method embodiments described below are merely illustrative.

step 1: and receiving traffic accident early warning information sent by the server so as to adjust the driving mode of the vehicle according to the early warning information.

In one example, the method may further comprise the steps of: and displaying the traffic accident early warning information so that a driver can adjust the driving mode of the vehicle according to the early warning information.

In another example, the traffic information broadcasting device is disposed in an autonomous vehicle, and the method may further include the steps of: and automatically adjusting the driving mode of the vehicle according to the traffic accident early warning information.

As can be seen from the above embodiments, the traffic information broadcasting method provided by the embodiments of the present application receives the traffic accident warning information sent by the server, so as to adjust the vehicle driving mode according to the traffic accident warning information; the processing mode can adjust the vehicle running mode in time; therefore, driving safety can be effectively improved.

Twenty-seventh embodiment

The application also provides an embodiment of a traffic information broadcasting device. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

Twenty-eighth embodiment

The present application further provides an embodiment of an electronic device. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

An electronic device of the present embodiment includes: a processor and a memory; the memory is used for storing a program for realizing the traffic information broadcasting method, and after the equipment is powered on and runs the program for realizing the traffic information broadcasting method through the processor, the following steps are executed: and receiving traffic accident early warning information sent by the server so as to adjust the driving mode of the vehicle according to the early warning information.

In the first embodiment, an event type determining method is provided, and correspondingly, an event occurrence mode determining system is also provided.

Twenty-ninth embodiment

Please refer to fig. 29, which is a diagram illustrating an embodiment of an event occurrence mode determining system according to the present application. Since the embodiment of the system corresponds to the embodiment of the method in the first embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the first embodiment. The system embodiments described below are merely illustrative.

The application provides an event occurrence pattern determination system, which comprises: a server 2901 and a client 2902. The client described in this application includes but is not limited to a mobile communication device, namely: the mobile phone or the smart phone also includes terminal devices such as a personal computer, a PAD, and an iPad.

Please refer to fig. 30, which is a schematic diagram illustrating an interaction of a device according to an embodiment of an event occurrence mode determining system provided in the present application. The server is used for acquiring at least one event text; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; and determining event occurrence mode information according to the event element information. Table 1 shows event occurrence pattern information and element information thereof of the present embodiment.

Table 1, the event occurrence mode information and the element information thereof are shown in table 1, if the service end determines that the traffic accident occurrence mode information is: the traffic lights are not installed at the intersections, and the straight-ahead accidents easily occur at the rush hour of commuting, so the types of the incident occurrence places are as follows: intersections without traffic lights installed; the event occurrence time is as follows: the peak time of going to work and off work; the event types are: and (5) a straight-going accident. If the server determines that the traffic accident occurrence mode information is as follows: the intersection at the position of the slope ascending first and the slope descending later is easy to have rear-end collision accidents caused by untimely braking in rainy and foggy weather, and the types of the accident occurrence places are as follows: a downhill crossing; the event occurrence climate is as follows: rain and fog weather; the event types are: rear-end accidents, etc.

The client is used for determining the target event element and sending an event occurrence mode information acquisition request aiming at the target event element to the server, wherein if the target event element is' climate: rain and fog weather "; correspondingly, the server is also used for receiving the request; determining event occurrence pattern information including the target event element based on predetermined event occurrence pattern information, such as an event occurrence model that is likely to occur in rainy and foggy weather, includes: "the vehicle is apt to take place the straight-going accident caused by brake untimely at the crossing of the downhill," the electric power enterprise apparatus A is apt to take place the electric leakage accident "etc.; returning event occurrence pattern information including the target event element to the client; and receiving event occurrence mode information comprising the target event element, so that a client user can perform decision control and the like according to the mode information.

In one example, the target event element is specific location information, such as specific locations: the method comprises the following steps that (1) at an intersection 056, a client sends an event occurrence mode information acquisition request aiming at the intersection to a server; correspondingly, the server side firstly determines that the place type of the intersection is the downhill intersection, and then determines event occurrence mode information related to the downhill intersection according to predetermined event occurrence mode information, such as 'rear-end collision accident caused by untimely braking easily happens to the intersection at the downhill in rainy and foggy weather', and a client side user can control the vehicle running speed according to the mode information; "during peak hours of work and work, the vehicles are easy to generate congestion events at the intersections of the downhill roads", and the client user can adjust the running roads of the vehicles according to the mode information.

As can be seen from the foregoing embodiments, the event occurrence mode determining system provided in the embodiments of the present application obtains at least one event text through the server; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information; receiving an event occurrence mode information acquisition request aiming at the target event element, which is sent by a client; determining event occurrence pattern information including the target event element; returning event occurrence pattern information including the target event element to the client; this processing manner makes it possible to determine event occurrence pattern information including a target event element; therefore, the efficiency and the accuracy of determining the event occurrence mode can be effectively improved.

Thirtieth embodiment

The application also provides an event occurrence mode determining method, and an executive body of the method comprises an event occurrence mode determining device. Since the method embodiment corresponds to the method embodiment of twenty-nine embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of twenty-nine embodiment. The method embodiments described below are merely illustrative.

step 1: acquiring at least one event text; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; and determining event occurrence mode information according to the event element information.

Step 2: and receiving an event occurrence mode information acquisition request aiming at the target event element, which is sent by the client.

And step 3: determining event occurrence pattern information including the target event element. And 4, step 4: and sending back the event occurrence mode information comprising the target event element to the client.

As can be seen from the foregoing embodiments, in the event occurrence mode determination method provided in the embodiments of the present application, at least one event text is obtained through a server; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information; receiving an event occurrence mode information acquisition request aiming at the target event element, which is sent by a client; determining event occurrence pattern information including the target event element; returning event occurrence pattern information including the target event element to the client; this processing manner makes it possible to determine event occurrence pattern information including a target event element; therefore, the efficiency and the accuracy of determining the event occurrence mode can be effectively improved.

Thirty-first embodiment

The application also provides an event occurrence mode determining device. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

Thirty-second embodiment

An electronic device of the present embodiment includes: a processor and a memory; the memory is used for storing a program for realizing the event occurrence mode determination method, and after the device is powered on and the program for realizing the event occurrence mode determination method is run by the processor, the following steps are executed: at least one event text is obtained in advance; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information; receiving an event occurrence mode information acquisition request aiming at a target event element, which is sent by a client; determining event occurrence pattern information including the target event element; and sending back the event occurrence mode information comprising the target event element to the client.

Thirty-third embodiment

The application also provides an embodiment of the method for determining the event occurrence mode, and an executive body of the method comprises an event occurrence mode determining device. Since the method embodiment corresponds to the method embodiment of thirty, the description is simple, and the relevant points can be referred to the partial description of thirty embodiment. The method embodiments described below are merely illustrative.

step S1: sending an event occurrence mode information acquisition request aiming at the target event element to a server;

step S2: and receiving the event occurrence mode information including the target event element returned by the server.

As can be seen from the foregoing embodiments, in the event occurrence mode determining method provided in the embodiments of the present application, an event occurrence mode information acquisition request for a target event element is sent to a server through a client, and the server acquires at least one event text in advance; determining words included in the text and event element types of the words aiming at each event text; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; determining event occurrence mode information according to the event element information; receiving an event occurrence mode information acquisition request aiming at the target event element, which is sent by a client; determining event occurrence pattern information including the target event element; returning event occurrence pattern information including the target event element to the client; this processing manner makes it possible to determine event occurrence pattern information including a target event element; therefore, the efficiency and the accuracy of determining the event occurrence mode can be effectively improved.

Thirty-fourth embodiment

The application also provides an embodiment of an event occurrence mode determination device. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

Thirty-fifth embodiment

An electronic device of the present embodiment includes: a processor and a memory; the memory is used for storing a program for realizing the event occurrence mode determination method, and after the device is powered on and the program for realizing the event occurrence mode determination method is run by the processor, the following steps are executed: sending an event occurrence mode information acquisition request aiming at the target event element to a server; and receiving the event occurrence mode information including the target event element returned by the server.

In the first embodiment, an event category determining method is provided, and correspondingly, an event element determining system is also provided.

Thirty-sixth embodiment

Please refer to fig. 31, which is a diagram illustrating an embodiment of an event element determining system according to the present application. Since the embodiment of the system corresponds to the embodiment of the method in the first embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the first embodiment. The system embodiments described below are merely illustrative.

An event element determination system provided by the present application includes: a server 3101 and a client 3202. The client described in this application includes but is not limited to a mobile communication device, namely: the mobile phone or the smart phone also includes terminal devices such as a personal computer, a PAD, and an iPad.

Please refer to fig. 32, which is a schematic device interaction diagram of an embodiment of an event element determining system according to the present application. The server is used for receiving an event element information determination request aiming at a target event text sent by a client; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; and returning the event factor information to the client.

And the client is used for sending the request to the server and receiving the event element information so that a client user can perform decision control and the like according to the element information.

As can be seen from the foregoing embodiments, the event element determining system provided in the embodiments of the present application receives, through a server, an event element information determining request for a target event text sent by a client; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; returning event factor information to the client; the processing mode enables the event element information of the target event text to be determined; therefore, the efficiency and the accuracy of determining the event elements can be effectively improved.

Thirty-seventh embodiment

The application also provides an event element determining method, and an executive body of the method comprises an event element determining device. Since the embodiment of the method corresponds to the embodiment of the system, the description is relatively simple, and the relevant points can be referred to the partial description of the embodiment of the system. The method embodiments described below are merely illustrative.

The event element determining method provided by the application comprises the following steps:

step 1: and receiving an event element information determination request aiming at the target event text sent by the client.

Step 2: determining words included in the text and event element types of the words.

And step 3: and extracting the event features comprising text semantic information and event element type information according to the word vector of the words and the element type vector of the event element types through the event feature extraction sub-network included in the event element determination model.

And 4, step 4: and determining event element information of the event text according to the event characteristics by using an event element prediction sub-network included in the event element determination model.

And 5: and returning the event factor information to the client. As can be seen from the foregoing embodiments, in the event element determining method provided in the embodiments of the present application, an event element information determining request for a target event text sent by a client is received by a server; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; returning event factor information to the client; the processing mode enables the event element information of the target event text to be determined; therefore, the efficiency and the accuracy of determining the event elements can be effectively improved.

In the above embodiment, an event element determining method is provided, and correspondingly, the application also provides an event element determining device. The apparatus corresponds to an embodiment of the method described above.

Thirty-eighth embodiment

The application also provides an event element determining device. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

The present application further provides an event element determination device including:

Thirty-ninth embodiment

An electronic device of the present embodiment includes: a processor and a memory; the memory is used for storing a program for realizing the event element determination method, and after the device is powered on and the program for realizing the event element determination method is run by the processor, the following steps are executed: receiving an event element information determination request aiming at a target event text sent by a client; determining words included in the text and event element types of the words; extracting sub-networks of event features included by the event element determination model, and extracting event features including text semantic information and event element type information according to a word vector of a word and an element type vector of the event element type; determining event element information of the event text according to the event characteristics through an event element prediction sub-network included in an event element determination model; and returning the event factor information to the client.

Fortieth embodiment

The application also provides an embodiment of the event element determining method, and an execution subject of the method comprises the event element determining device. Since the embodiment of the method corresponds to the embodiment of the system, the description is relatively simple, and the relevant points can be referred to the partial description of the embodiment of the system. The method embodiments described below are merely illustrative.

step S1: sending an event element information determination request aiming at a target event text to a server;

step S2: and receiving the event factor information returned by the server.

As can be seen from the foregoing embodiments, in the event element determining method provided in the embodiments of the present application, an event element information determining request for a target event text is sent to a server through a client; receiving event element information returned by the server; the processing mode enables the event element information of the target event text to be determined; therefore, the efficiency and the accuracy of determining the event elements can be effectively improved.

Forty-first embodiment

The application also provides an embodiment of an event element determining device. Since the apparatus embodiments are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for relevant points. The device embodiments described below are merely illustrative.

Forty-second embodiment

An electronic device of the present embodiment includes: a processor and a memory; the memory is used for storing a program for realizing the event element determination method, and after the device is powered on and the program for realizing the event element determination method is run by the processor, the following steps are executed: sending an event element information determination request aiming at a target event text to a server; and receiving the event factor information returned by the server.

Although the present application has been described with reference to the preferred embodiments, it is not intended to limit the present application, and those skilled in the art can make variations and modifications without departing from the spirit and scope of the present application, therefore, the scope of the present application should be determined by the claims that follow.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.

1. Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include non-transitory computer readable media (transient media), such as modulated data signals and carrier waves.

2. As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

Claims

1. A method for determining an equipment failure occurrence mode is characterized by comprising the following steps:

acquiring a text related to equipment failure;

2. The method of claim 1, wherein the equipment failure element type is determined by:

3. The method according to claim 1, wherein the extracting of the device failure feature including text semantic information and device failure element type information from the word vector of words and the element type vector of the device failure element type by the device failure feature extraction sub-network included by the failure element determination model includes:

4. An apparatus for determining an occurrence mode of a device failure, comprising:

5. An electronic device, comprising:

a processor; and

6. A method for determining an event occurrence pattern, comprising:

acquiring an event related text;

determining words included in the text and event element types of the words;

7. The method of claim 6,

the event occurrence pattern information includes: patterns of event occurrences related to public transportation facilities;

the method further comprises the following steps:

8. An event occurrence pattern determination apparatus, comprising:

an event text acquisition unit for acquiring an event-related text;

9. An electronic device, comprising:

a processor; and

10. A method for determining an occurrence mode of an enterprise acquisition event is characterized by comprising the following steps:

acquiring a text related to an enterprise purchase event;

11. An apparatus for determining an occurrence pattern of an enterprise acquisition event, comprising:

12. An electronic device, comprising:

a processor; and

13. An event category determination method, comprising:

14. An event category determination device, comprising:

15. An electronic device, comprising:

a processor; and

16. A method for constructing an event element type prediction model is characterized by comprising the following steps:

17. An event element type prediction model construction device, comprising:

18. An electronic device, comprising:

a processor; and

19. A method for constructing an event classification model is characterized by comprising the following steps:

20. An event classification model construction device, comprising:

21. An electronic device, comprising:

a processor; and

22. A method for determining a traffic accident occurrence pattern, comprising:

acquiring a text related to a traffic accident;

23. A traffic accident occurrence pattern determination apparatus, comprising:

24. An electronic device, comprising:

a processor; and

25. A traffic information broadcasting system, comprising:

26. A traffic information broadcasting method is characterized by comprising the following steps:

determining the position information of a vehicle to be early-warned;

27. A traffic information broadcasting method is characterized by comprising the following steps:

28. A traffic information broadcasting device, comprising:

29. A traffic information broadcasting device, comprising:

30. An electronic device, comprising:

a processor; and

31. An electronic device, comprising:

a processor; and

32. An event occurrence pattern determination system, comprising:

33. A method for determining an event occurrence pattern, comprising:

34. A method for determining an event occurrence pattern, comprising:

35. An event occurrence pattern determination apparatus, comprising:

36. An event occurrence pattern determination apparatus, comprising:

37. An electronic device, comprising:

a processor; and

38. An electronic device, comprising:

a processor; and

39. An event element determination system, comprising:

40. An event element determination method, comprising:

determining words included in the text and event element types of the words;

and returning the event factor information to the client.

41. An event element determination method, comprising:

and receiving the event factor information returned by the server.

42. An event element determination device, comprising:

43. An event element determination device, comprising:

44. An electronic device, comprising:

a processor; and

45. An electronic device, comprising:

a processor; and