CN112784034A

CN112784034A - Abstract generation method and device and computer equipment

Info

Publication number: CN112784034A
Application number: CN201911060461.7A
Authority: CN
Inventors: 张雅婷; 周鑫; 孙常龙; 张琼; 司罗
Original assignee: Alibaba Group Holding Ltd
Current assignee: Alibaba Group Holding Ltd
Priority date: 2019-11-01
Filing date: 2019-11-01
Publication date: 2021-05-11

Abstract

The application discloses a method and a device for generating an abstract and computer equipment. Wherein, the method comprises the following steps: acquiring character information of a court trial to be detected; determining an element set of each role in the to-be-detected court trial text information based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the dispute key points of each role; inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus; extracting summary information of the sentence where the at least one dispute focus is located; and outputting the summary information. The method and the device solve the technical problem that manual combing of dispute focuses in court trial records influences the case trial efficiency.

Description

Abstract generation method and device and computer equipment

Technical Field

The present application relates to the field of computers, and in particular, to a method and an apparatus for generating an abstract, and a computer device.

Background

In the related art, the court trial record is usually very long, and about 1-3 ten thousand characters of the court trial record can be generated in 1 or 2 hours of court trial. This results in that the court officer needs to spend a lot of time on the dispute focus combing of the case in the court trial or after the court trial is finished, and the efficiency of the case management is affected.

In view of the above problems, no effective solution has been proposed.

Disclosure of Invention

The embodiment of the application provides a summary generation method, a summary generation device and computer equipment, and aims to at least solve the technical problem that manual combing and disputing focuses in court trial records influence the case trial efficiency.

According to an aspect of an embodiment of the present application, there is provided a digest generation method, including: acquiring character information of a court trial to be detected; determining an element set of each role in the to-be-detected court trial text information based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the dispute key points of each role; inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus; extracting summary information of the sentence where the at least one dispute focus is located; and outputting the summary information.

According to an aspect of an embodiment of the present application, there is provided a method for recognizing a statement type, including: acquiring a second target sentence in the text information of the court trial to be detected; determining a dispute focus statement from the second target statement; determining the similarity of the dispute focus statement and each sample dispute focus statement in the dispute focus statement set to obtain a plurality of similarities; and taking the type corresponding to the sample dispute focus statement corresponding to the maximum value in the similarity as the type of the second target statement.

According to an aspect of an embodiment of the present application, there is provided a digest generation method, including: acquiring character information of a court trial to be detected; determining an element set of each role in the to-be-detected court trial text information based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the dispute key points of each role; determining at least one dispute focus based on elements of the plurality of element sets; extracting summary information of the sentence where the at least one dispute focus is located; and outputting the summary information.

According to another aspect of the embodiments of the present application, there is also provided a digest generation apparatus, including: the acquisition module is used for acquiring the character information of the court trial to be detected; the determining module is used for determining an element set of each role in the court trial text information to be detected based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the key points of the disputes of each role; the analysis module is used for inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus; the extracting module is used for extracting the summary information of the statement where the at least one dispute focus is located; and the output module is used for outputting the summary information.

According to another aspect of the embodiments of the present application, there is also provided a computer device, including: a processor; and a memory coupled to the processor for providing instructions to the processor for processing the following processing steps: acquiring character information of a court trial to be detected; determining an element set of each role in the to-be-detected court trial text information based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the dispute key points of each role; inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus; extracting summary information of the sentence where the at least one dispute focus is located; and outputting the summary information.

In the embodiment of the application, acquiring the text information of the court trial to be detected; determining an element set of each role in the to-be-detected court trial text information based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the dispute key points of each role; inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus; extracting summary information of the sentence where the at least one dispute focus is located; the mode of outputting the abstract information automatically identifies the dispute focus in the court trial text information and outputs the abstract information, thereby realizing the technical effects of reducing the time for a judge to obtain the dispute focus and the case key points and improving the case trial efficiency, and further solving the technical problem that the manual combing of the dispute focus in the court trial records influences the case trial efficiency.

Drawings

The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:

fig. 1 is a block diagram of a hardware configuration of a computer terminal according to an embodiment of the present application;

FIG. 2 is a schematic flow chart diagram illustrating an alternative summary generation method according to an embodiment of the present application;

FIG. 3 is an alternative legal knowledge element map according to an embodiment of the present application;

FIG. 4a is a schematic diagram of a summarization principle according to an embodiment of the present application;

FIG. 4b is a flow chart illustrating an alternative sentence type identification method according to an embodiment of the present application;

FIG. 5 is a flowchart illustrating a method for generating a summary according to an embodiment of the present application;

FIG. 6 is a schematic structural diagram of an alternative summary generation apparatus according to an embodiment of the present application;

fig. 7 is a schematic structural diagram of an alternative computer device according to an embodiment of the present application.

Detailed Description

In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only partial embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

It should be noted that the terms "first," "second," and the like in the description and claims of this application and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

First, some terms or terms appearing in the description of the embodiments of the present application are applicable to the following explanations:

the voice recognition technology comprises the following steps: also known as ASR (Automatic Speech Recognition), which aims at converting the vocabulary content in human Speech into computer-readable input, such as keystrokes, binary codes or character sequences.

The dispute focus is as follows: the focus of dispute is the major problem to be solved after disputes between parties have occurred, which is first of all a problem, specifically a major problem involving facts, evidence, legal provisions, liability, etc. that cause disputes.

Example 1

There is also provided, in accordance with an embodiment of the present application, a method of generating a summary, it being noted that the steps illustrated in the flowchart of the drawings may be performed in a computer system such as a set of computer-executable instructions and that, although a logical order is illustrated in the flowchart, in some cases the steps illustrated or described may be performed in an order different than here.

The method provided by the first embodiment of the present application may be executed in a mobile terminal, a computer terminal, or a similar computing device. Fig. 1 shows a hardware configuration block diagram of a computer terminal (or mobile device) for implementing the digest generation method. As shown in fig. 1, the computer terminal 10 (or mobile device 10) may include one or more (shown as 102a, 102b, … …, 102 n) processors 102 (the processors 102 may include, but are not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission module 106 for communication functions. Besides, the method can also comprise the following steps: a display, an input/output interface (I/O interface), a Universal Serial Bus (USB) port (which may be included as one of the ports of the I/O interface), a network interface, a power source, and/or a camera. It will be understood by those skilled in the art that the structure shown in fig. 1 is only an illustration and is not intended to limit the structure of the electronic device. For example, the computer terminal 10 may also include more or fewer components than shown in FIG. 1, or have a different configuration than shown in FIG. 1.

It should be noted that the one or more processors 102 and/or other data processing circuitry described above may be referred to generally herein as "data processing circuitry". The data processing circuitry may be embodied in whole or in part in software, hardware, firmware, or any combination thereof. Further, the data processing circuit may be a single stand-alone processing module, or incorporated in whole or in part into any of the other elements in the computer terminal 10 (or mobile device). As referred to in the embodiments of the application, the data processing circuit acts as a processor control (e.g. selection of a variable resistance termination path connected to the interface).

The memory 104 may be used to store software programs and modules of application software, such as program instructions/data storage devices corresponding to the summary generation method in the embodiment of the present application, and the processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, implementing the vulnerability detection method of the application program. The memory 104 may include high speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include memory located remotely from the processor 102, which may be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The transmission module 106 is used to receive or transmit data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission module 106 includes a Network adapter (NIC) that can be connected to other Network devices through a base station to communicate with the internet. In one example, the transmission module 106 may be a Radio Frequency (RF) module, which is used for communicating with the internet in a wireless manner.

The display may be, for example, a touch screen type Liquid Crystal Display (LCD) that may enable a user to interact with a user interface of the computer terminal 10 (or mobile device).

Under the above operating environment, the present application provides a summary generation method as shown in fig. 2. Fig. 2 is a flowchart of a summary generation method according to a first embodiment of the present application, where the method includes the following processing flows:

step S202, acquiring character information of a court trial to be detected;

specifically, the court trial text information to be detected can be voice debate information or text debate information of an original advertiser and an advertised advertiser in the court case examination process, and can also be text debate information before court trial;

in some optional embodiments of the present application, the method further needs to perform the following steps: detecting current voice information; and converting the voice information into character information, and using the character information as the character information of the court trial to be detected. The voice acquisition device for acquiring the voice information can be installed at a preset position of the court, and the real-time voice-to-text transcription device is arranged and operates in the court trial process in real time. The conversation of the party and the judge can carry out real-time transcription of voice to characters by supporting multi-role ASR, the minimum transcription unit is the character level, and the transcription result can be dynamically updated according to the context, so that a better transcription effect is achieved. And after ASR, accessing the court trial transcription personalized transcription system, and performing smooth processing on the ASR result, wherein the smooth processing comprises the steps of eliminating sentence break errors, deleting repeated spoken language, eliminating entity recognition errors, eliminating legal phrase recognition errors and the like. Wherein the entities include: name of person, place name, organization name, etc.

Step S204, determining an element set of each role in the court trial text information to be detected based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the dispute key points of each role;

in some optional embodiments of the present application, the legal knowledge element map is a map of legal elements involved in a legal incident, for example, a principal dispute, as shown in fig. 3, and the legal knowledge element map includes: loan agreement, payment, written agreement associated with the loan agreement or electronic agreement, oral agreement; and delivery credentials, cash (ticket) delivery, etc. related to the payment of the money.

Specifically, the elements in the element set of each role may be determined by the following information: character debate information before court trial and debate information in the court trial process. There are disputes of different roles in the court trial process, and the element set of the original advertiser and the advertised advertiser can be automatically identified based on the legal knowledge element map. The legal knowledge element map can be a hierarchical tree, and each role can be endowed with a hierarchical element set through prediction of each sentence element of one role. Here, a hierarchical element set can be understood as a plurality of hierarchical element paths, for example, "principal-loan consensus-signature" is an element path, and one element path set, i.e., K element paths, is predicted for N sentences of each role. The hierarchical element set is obtained by a multi-label hierarchical classification model, wherein N words are input, and K element paths are output.

Step S206, inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus; wherein, this dispute focus model is obtained for training through multiunit data, all includes in every group data in this multiunit data: a sample element and a label identifying the sample element as a point of dispute focus.

Specifically, when the input of the dispute focus model is an element in the hierarchical element set of the multi-role, at least one dispute focus corresponding to the multi-role is output.

And step S208, extracting summary information of the statement where the at least one dispute focus is located. In some optional embodiments of the present application, the summary information is summary information that may reflect the dispute focus of the case.

Step S210, outputting the summary information;

in some optional embodiments of the present application, the method further needs to perform the following steps: determining the sentence where the at least one dispute focus is located to obtain a first target sentence; determining a sentence vector of the first target sentence based on the semantic representation of the first target sentence and the element representation of the first target sentence in the legal element map; obtaining similarity of statement vectors of the first target statement and each sample dispute focus statement vector in a dispute focus statement set to obtain a plurality of similarities; determining a dispute focus type to which the first target statement belongs based on the plurality of similarities. The sample dispute focus statement may be a sample dispute focus statement obtained based on court trial information in a court trial process of a similar historical case.

Specifically, determining the type of dispute focus to which the first target sentence belongs based on the plurality of similarities may be implemented by the following processing steps: and comparing the sizes of the plurality of similarities, and taking the dispute focus type to which the dispute focus statement corresponding to the maximum value in the plurality of similarities belongs as the dispute focus type of the first target statement.

In order to avoid the problem that the similarity between the first target statement and the sample dispute focus statement is low, which results in inaccurate generated summary information, the method may further include the following steps: determining a maximum value of the plurality of similarities; comparing the maximum value to a first threshold value; when the comparison result indicates that the maximum value is larger than the first threshold value, determining to extract summary information of the statement where the at least one dispute focus is located; and when the comparison result indicates that the maximum value is smaller than the first threshold value, refusing to extract summary information of the statement where the at least one dispute focus is located.

Before abstract information of a sentence where at least one dispute focus is located is extracted, after similarity between a sentence vector of a first target sentence and vectors of all sample dispute focus sentences is determined, the abstract information of the sentence where the at least one dispute focus is located is determined to be extracted when the maximum value of the similarity exceeds a set first threshold value. By adopting the process, the accuracy of abstract information extraction is ensured.

In order to facilitate a judge to visually acquire the dispute between the defendant and the defendant of the current case through the summary information, when the summary information is output, the method further needs to execute the following steps: and outputting the dispute focus type to which the first target statement belongs, the role information corresponding to the first target statement and the association relationship between the role information of the dispute focus type and the summary information. The dispute focus type of the first target statement, the role information corresponding to the first target statement and the display of the association relationship between the role information of the dispute focus type and the summary information are beneficial for a judge to quickly acquire the dispute progress of the defendant and the original defender and better grasp the time.

Specifically, the output dispute focus type is represented by a dispute focus type identifier, and different dispute focus type identifiers are in different display forms.

Optionally, the dispute focus type identifier may be an arabic numeral number, or a combination of a numeral number and a letter (e.g., d1, d2, d3 · dn, etc.), and the dispute focus type identifiers may be displayed in different colors or in different fonts.

In some optional embodiments of the present application, the method further needs to perform the following steps: and highlighting the sentence where the at least one dispute focus is in the to-be-detected court trial text information. The highlighting mode may be highlighting. For example, the dispute focus automatically identified by the dispute focus model may be represented in different colors, and the sentence in which the corresponding extracted dispute focus is located may also be highlighted in the color of the corresponding dispute focus. The summary may also be a listing of multi-role conversations surrounding the focus of the dispute.

When obtaining the similarity between the statement vector of the first target statement and each sample dispute focus statement vector in the dispute focus statement set, the similarity may be determined in a machine learning manner, where the machine learning may be a regression model, for example:

inputting the text information corresponding to each role in the court trial character information and each element in the legal knowledge map into a first-stage learning model respectively to perform representation learning of a single information source to obtain characteristic information of each information source; determining each role, and performing characterization joint processing on the text information corresponding to each role and each element to obtain the speech content characterization of each role; inputting the representation of the speaking content of each role into a second-level learning model to obtain relevant dialogue information; and inputting the dialog information with the correlation and the dispute focus statements in the dispute focus statement set to an attention model for analysis to obtain the similarity. Specifically, the method comprises the following steps:

first, a representation (representation) of a court trial sentence is constructed as a semantic representation and an element representation based on an element map. Semantic representation, namely encoding the semantics of the sentence through a pre-trained sentence encoder (sentencecoder) to form an embedding vector representation; the element representation is represented by mapping the sentence to an element map and using an element map vector. The concatenation of the two representations serves as a representation of the sentence. For example, as shown in fig. 4a, text information (r1, r2, r3, · · rm) corresponding to a plurality of characters in court trial text information and each word and word (W1, W2, · · Wn, Wn) in the knowledge graph information are respectively input into a first-level learning model (e.g., GRU model) to perform characterization learning of a single information source, and then the characterizations of the results of the characterization learning of each information source (characters, texts and each word (i.e., element) in the knowledge graph) are combined (concat) to serve as the characterization of the current speaking content of the character, i.e., the representation of the court trial sentence is obtained;

the joined tokens are then input to a second level learning model (e.g., a GRU model) that learns the token-level tokens between adjacent dialogs, i.e., learns the correlations between dialogs. Weighting each sentence in the court trial text information by utilizing an attention layer on the basis of an output result of the second-level learning model and the obtained dispute focus so as to calculate the similarity between the sentence in the court trial text information and the sentence in the dispute focus sentence set, thereby classifying the sentences in the court trial text information and obtaining a classification label of each sentence.

The above calculation process of the similarity can be understood as follows: performing a sentence classification subtask around the dispute focus: this subtask may be understood as dynamically calculating the similarity between the current sentence and the dispute focus, with the most similar dispute focus as the classification label for the sentence.

Based on the above scenario, after determining the similarity, in order to obtain the abstract (i.e. the important sentence), an abstract extraction sub-task may be performed: the subtask may be the process of calculating a regression, i.e. whether the current sentence should be extracted or not. It should be noted that the process of constructing the identification of the court trial sentences, and executing the grouping classification subtask surrounding the dispute focus and the abstract extraction subtask is a multi-task learning process, that is, all the parameters are updated simultaneously in the learning process, so that the requirement that the learned sentences meet the sentence classification task and the abstract extraction task is met.

It should be noted that, for simplicity of description, the above-mentioned method embodiments are described as a series of acts or combination of acts, but those skilled in the art will recognize that the present application is not limited by the order of acts described, as some steps may occur in other orders or concurrently depending on the application. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required in this application.

Through the above description of the embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but the former is a better implementation mode in many cases. Based on such understanding, the technical solutions of the present application may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present application.

Example 2

According to the embodiment of the present application, there is also provided a sentence type identification method, and fig. 4b is a flowchart of the sentence type identification method according to embodiment 2 of the present application, and as shown in fig. 4b, the method includes the following processing flows of steps S402-S408:

step S402, acquiring a second target sentence in the character information of the court trial to be detected;

specifically, the court trial text information to be detected can be voice debate information or text debate information of an original advertiser and an advertiser in the court case examination process, also can be text debate information before court trial, and the second target statement is a statement of the debate information;

in some optional embodiments of the present application, a voice collecting device for obtaining voice information may be installed at a preset position in a court, and a real-time voice-to-text transcribing device may be provided, where the real-time voice-to-text transcribing device operates in the court trial process in real time. The conversation of the party and the judge can carry out real-time transcription of voice to characters by supporting multi-role ASR, the minimum transcription unit is the character level, and the transcription result can be dynamically updated according to the context, so that a better transcription effect is achieved. And after ASR, accessing the court trial transcription personalized transcription system, and performing smooth processing on the ASR result, wherein the smooth processing comprises the steps of eliminating sentence break errors, deleting repeated spoken language, eliminating entity recognition errors, eliminating legal phrase recognition errors and the like. Wherein the entities include: name of person, place name, organization name, etc.

The second target sentence may also be a smoothed debate sentence.

Specifically, an element set of each role in the second target sentence needs to be determined based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the debate points of each role;

step S404, determining a dispute focus statement from the second target statement;

in some optional embodiments of the present application, the legal knowledge element map is a map of legal elements involved in a legal incident, for example, a principal dispute, as shown in fig. 3, and the legal knowledge element map includes: a loan agreement, a payment, a written agreement or an electronic agreement, a verbal agreement, a proof of delivery, a cash (instrument) delivery, etc.

Specifically, the elements in the element set of each role may be determined by the following information: character debate information before court trial and debate information in the court trial process. There are disputes of different roles in the court trial process, and the element set of the original advertiser and the advertised advertiser can be automatically identified based on the legal knowledge element map. The legal knowledge element map can be a hierarchical tree, and each role can be endowed with a hierarchical element set through prediction of each sentence element of one role.

Inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus;

wherein, this dispute focus model is obtained for training through multiunit data, all includes in every group data in this multiunit data: a sample element and a label identifying the sample element as a point of dispute focus.

Specifically, when the input of the dispute focus model is an element in the hierarchical element set of the multi-role, at least one dispute focus corresponding to the multi-role is output. Specifically, the dispute focus automatically identified by the dispute focus model may be represented in different colors, and the corresponding sentence in which the extracted dispute focus is located is also highlighted in the color of the corresponding dispute focus. The summary may also be a listing of multi-role conversations surrounding the focus of the dispute. The second target statement at the dispute focus is the dispute focus statement.

Step S406, determining the similarity of the dispute focus statement and each sample dispute focus statement in the dispute focus statement set to obtain a plurality of similarities;

step S408, using the type corresponding to the sample dispute focus statement corresponding to the maximum value in the multiple similarities as the type of the second target statement.

Example 3

According to an embodiment of the present application, there is also provided a digest generation method, as shown in fig. 5, fig. 5 is a flowchart of a digest generation method according to embodiment 3 of the present application, and the method includes the following processing flows:

step S502, acquiring character information of a court trial to be detected;

Step S504, determining an element set of each role in the court trial text information to be detected based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the dispute key points of each role;

Step S506, determining at least one dispute focus based on the elements in the plurality of element sets;

optionally, determining at least one dispute focus based on elements of the plurality of element sets may be performed by: inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus; wherein, this dispute focus model is obtained for training through multiunit data, all includes in every group data in this multiunit data: a sample element and a label identifying the sample element as a point of dispute focus.

Specifically, the dispute focus automatically identified by the dispute focus model may be represented in different colors, and the corresponding sentence in which the extracted dispute focus is located is also highlighted in the color of the corresponding dispute focus. The summary may also be a listing of multi-role conversations surrounding the focus of the dispute.

Step S508, abstract information of the statement where the at least one dispute focus is located is extracted; in some optional embodiments of the present application, the summary information is summary information that can reflect the dispute focus of the case;

and step S510, outputting the summary information.

In some optional embodiments of the present application, the method further needs to perform the following steps: determining the sentence where the at least one dispute focus is located to obtain a first target sentence; determining a sentence vector of the first target sentence based on the semantic representation of the first target sentence and the element representation of the first target sentence in the legal element map; obtaining similarity of statement vectors of the first target statement and each sample dispute focus statement vector in a dispute focus statement set to obtain a plurality of similarities; determining a dispute focus type to which the first target statement belongs based on the plurality of similarities.

The sample dispute focus statement may be a sample dispute focus statement obtained based on court trial information in a court trial process of a similar historical case.

Before abstract information of a statement where at least one dispute focus is located is extracted, after similarity between a statement vector of a first target statement and vectors of various sample dispute focus statements is determined, when the maximum value of the similarity exceeds a set first threshold value, the abstract information of the statement where the at least one dispute focus is located is determined to be extracted; by adopting the process, the accuracy of abstract information extraction is ensured.

Alternatively, the dispute focus type identifier may be an arabic number of the dispute focus, and the display form of different dispute focus type identifiers may be different colors or different fonts.

In some optional embodiments of the present application, the method further needs to perform the following steps: and highlighting the sentence where the at least one dispute focus is in the to-be-detected court trial text information.

The highlighting mode may be highlighting.

Example 4

According to an embodiment of the present application, there is also provided a digest generation apparatus for implementing the digest generation method, as shown in fig. 6, the apparatus includes: the device comprises an acquisition module 62, a determination module 64, an analysis module 66, an extraction module 68 and an output module 610; wherein:

the acquisition module 62 is used for acquiring the character information of the court trial to be detected;

a determining module 64, configured to determine, based on a legal knowledge element map, an element set of each role in the court trial text information to be detected, so as to obtain multiple element sets, where an element in the element set is used to reflect a dispute point of each role;

the analysis module 66 is configured to input the elements in the multiple element sets into a dispute focus model for analysis, so as to obtain at least one dispute focus;

an extracting module 68, configured to extract summary information of the statement at which the at least one dispute focus is located;

an output module 610, configured to output the summary information.

The determining module 64 is further configured to determine a statement where the at least one dispute focus is located, so as to obtain a first target statement; determining a sentence vector of the first target sentence based on the semantic representation of the first target sentence and the element representation of the first target sentence in the legal element map; obtaining similarity of statement vectors of the first target statement and each sample dispute focus statement vector in a dispute focus statement set to obtain a plurality of similarities; determining a dispute focus type to which the first target statement belongs based on the plurality of similarities.

The determining module 64 is further configured to compare the magnitudes of the multiple similarities, and use a dispute focus type to which the dispute focus statement corresponding to the maximum value in the multiple similarities belongs as the dispute focus type of the first target statement.

The determination module 64 is further configured to: determining a maximum value of the plurality of similarities; comparing the maximum value to a first threshold value; when the comparison result indicates that the maximum value is larger than the first threshold value, determining to extract summary information of the statement where the at least one dispute focus is located; and when the comparison result indicates that the maximum value is smaller than the first threshold value, refusing to extract summary information of the statement where the at least one dispute focus is located.

The apparatus further includes a summarization module, configured to output, when the output module 610 outputs the summary information, a dispute focus type to which the first target statement belongs, role information corresponding to the first target statement, and an association relationship between the role information of the dispute focus type and the summary information.

The dispute focus type is represented by a dispute focus type identifier, and different dispute focus type identifiers are in different display forms.

The device also comprises a display module which is used for highlighting the sentence where the at least one dispute focus is located in the to-be-detected court trial text information.

The device also comprises a detection module used for detecting the current voice information; and converting the voice information into character information, and using the character information as the character information of the court trial to be detected. Wherein, the elements in the element set of each role are determined by the following information: character debate information before court trial and debate information in the court trial process.

It should be noted here that the obtaining module 62, the determining module 64, the analyzing module 66, the extracting module 68, and the outputting module 610 correspond to steps S202 to S210 in embodiment 1, and the two modules are the same as the corresponding steps in the implementation example and the application scenario, but are not limited to the disclosure in the first embodiment. It should be noted that the modules described above as part of the apparatus may be run in the computer terminal 10 provided in the first embodiment.

Example 5

Embodiments of the present application may provide a computer device that may be any one of a group of computer devices. Optionally, in this embodiment, the computer device may also be replaced with a terminal device such as a mobile terminal.

Optionally, in this embodiment, the computer device may be located in at least one network device of a plurality of network devices of a computer network.

In this embodiment, the computer device may execute program codes of the following steps in the digest generation method of the application program:

acquiring character information of a court trial to be detected; determining an element set of each role in the to-be-detected court trial text information based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the dispute key points of each role; inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus; extracting summary information of the sentence where the at least one dispute focus is located; and outputting the summary information.

Optionally, fig. 7 is a block diagram of a computer device according to an embodiment of the present application. As shown in fig. 7, the computer device 700 may include: one or more (only one shown) processors 702, memory 704.

The memory 704 may be used to store software programs and modules, such as program instructions/modules corresponding to the summary generation method and apparatus in the embodiments of the present application, and the processor 702 executes various functional applications and data processing by running the software programs and modules stored in the memory 704, so as to implement the summary generation method described above. The memory 704 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 704 may further include memory located remotely from the processor 702, which may be connected to the terminal over a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The processor 702 may call the information and applications stored in the memory 704 through the transmission module to perform the following steps:

Optionally, the processor 702 may further execute the program code of the following steps: determining the sentence where the at least one dispute focus is located to obtain a first target sentence; determining a sentence vector of the first target sentence based on the semantic representation of the first target sentence and the element representation of the first target sentence in the legal element map; obtaining similarity of statement vectors of the first target statement and each sample dispute focus statement vector in a dispute focus statement set to obtain a plurality of similarities; determining a dispute focus type to which the first target statement belongs based on the plurality of similarities.

Optionally, the processor 702 may further execute the program code of the following steps: and comparing the sizes of the plurality of similarities, and taking the dispute focus type to which the dispute focus statement corresponding to the maximum value in the plurality of similarities belongs as the dispute focus type of the first target statement.

Optionally, the processor 702 may further execute the program code of the following steps: determining a maximum value of the plurality of similarities; comparing the maximum value to a first threshold value; when the comparison result indicates that the maximum value is larger than the first threshold value, determining to extract summary information of the statement where the at least one dispute focus is located; and when the comparison result indicates that the maximum value is smaller than the first threshold value, refusing to extract summary information of the statement where the at least one dispute focus is located.

Optionally, the processor 702 may further execute the program code of the following steps: and outputting the dispute focus type to which the first target statement belongs, the role information corresponding to the first target statement and the association relationship between the role information of the dispute focus type and the summary information.

Optionally, the output dispute focus type is represented by a dispute focus type identifier, and different dispute focus type identifiers are displayed in different forms.

Optionally, the processor 702 may further execute the program code of the following steps: and highlighting the sentence where the at least one dispute focus is in the to-be-detected court trial text information.

Optionally, the processor 702 may further execute the program code of the following steps: detecting current voice information; and converting the voice information into character information, and using the character information as the character information of the court trial to be detected.

Optionally, the elements in the element set of each role are determined by the following information: character debate information before court trial and debate information in the court trial process.

It will be understood by those skilled in the art that the structure shown in fig. 7 is only an illustration, and the computer device 700 may also be a terminal device such as a smart phone (e.g., an Android phone, an iOS phone, etc.), a tablet computer, a palmtop computer, a Mobile Internet Device (MID), a PAD, etc. Fig. 7 is a diagram illustrating a structure of the electronic device. For example, computer device 700 may also include more or fewer components (e.g., network interfaces, display devices, etc.) than shown in FIG. 7, or have a different configuration than shown in FIG. 7.

Those skilled in the art will appreciate that all or part of the steps in the methods of the above embodiments may be implemented by a program instructing hardware associated with the terminal device, where the program may be stored in a computer-readable storage medium, and the storage medium may include: flash disks, Read-Only memories (ROMs), Random Access Memories (RAMs), magnetic or optical disks, and the like.

Embodiments of the present application also provide a storage medium. Optionally, in this embodiment, the storage medium may be configured to store a program code executed by the digest generation method provided in the first embodiment.

Optionally, in this embodiment, the storage medium may be located in any one of computer terminals in a computer terminal group in a computer network, or in any one of mobile terminals in a mobile terminal group.

Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: acquiring character information of a court trial to be detected; determining an element set of each role in the to-be-detected court trial text information based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the dispute key points of each role; inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus; extracting summary information of the sentence where the at least one dispute focus is located; and outputting the summary information.

The above-mentioned serial numbers of the embodiments of the present application are merely for description and do not represent the merits of the embodiments.

In the above embodiments of the present application, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.

In the embodiments provided in the present application, it should be understood that the disclosed technology can be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one type of division of logical functions, and there may be other divisions when actually implemented, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, units or modules, and may be in an electrical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a removable hard disk, a magnetic or optical disk, and other various media capable of storing program codes.

The foregoing is only a preferred embodiment of the present application and it should be noted that those skilled in the art can make several improvements and modifications without departing from the principle of the present application, and these improvements and modifications should also be considered as the protection scope of the present application.

Claims

1. A method for generating a summary, comprising:

acquiring character information of a court trial to be detected;

determining an element set of each role in the to-be-detected court trial text information based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the dispute key points of each role;

extracting summary information of the sentence where the at least one dispute focus is located;

and outputting the summary information.

2. The method of claim 1, further comprising:

determining the sentence where the at least one dispute focus is located to obtain a first target sentence;

determining a sentence vector of the first target sentence based on the semantic representation of the first target sentence and the element representation of the first target sentence in the legal element map;

obtaining similarity of statement vectors of the first target statement and each sample dispute focus statement vector in a dispute focus statement set to obtain a plurality of similarities;

determining a dispute focus type to which the first target statement belongs based on the plurality of similarities.

3. The method of claim 2, wherein determining the type of dispute focus to which the first target statement belongs based on the plurality of similarities comprises:

and comparing the sizes of the plurality of similarities, and taking the dispute focus type to which the dispute focus statement corresponding to the maximum value in the plurality of similarities belongs as the dispute focus type of the first target statement.

4. The method of claim 2, further comprising:

determining a maximum value of the plurality of similarities;

comparing the maximum value to a first threshold value;

when the comparison result indicates that the maximum value is larger than the first threshold value, determining to extract summary information of the statement where the at least one dispute focus is located; and when the comparison result indicates that the maximum value is smaller than the first threshold value, refusing to extract summary information of the statement where the at least one dispute focus is located.

5. The method of claim 2, wherein when outputting the summary information, the method further comprises: and outputting the dispute focus type to which the first target statement belongs, the role information corresponding to the first target statement and the association relationship between the role information of the dispute focus type and the summary information.

6. The method of claim 5, wherein the outputted dispute focus type is represented by a dispute focus type identifier, and wherein different dispute focus type identifiers are displayed in different forms.

7. The method of claim 2, wherein obtaining similarity between the statement vector of the first target statement and each sample dispute focus statement vector in the set of dispute focus statements comprises:

inputting the text information corresponding to each role in the court trial character information and each element in the legal knowledge map into a first-stage learning model respectively to perform representation learning of a single information source to obtain characteristic information of each information source;

determining each role, and performing characterization joint processing on the text information corresponding to each role and each element to obtain the speech content characterization of each role;

inputting the representation of the speaking content of each role into a second-level learning model to obtain relevant dialogue information;

and inputting the dialog information with the correlation and the dispute focus statements in the dispute focus statement set to an attention model for analysis to obtain the similarity.

8. The method of claim 1, further comprising:

and highlighting the sentence where the at least one dispute focus is in the to-be-detected court trial text information.

9. The method according to any one of claims 1 to 7, further comprising:

detecting current voice information; and converting the voice information into character information, and using the character information as the character information of the court trial to be detected.

10. The method according to any one of claims 1 to 8, wherein the elements in the set of elements of each role are determined by:

character debate information before court trial and debate information in the court trial process.

11. A sentence type identification method is characterized by comprising the following steps:

acquiring a second target sentence in the text information of the court trial to be detected;

determining a dispute focus statement from the second target statement;

determining the similarity of the dispute focus statement and each sample dispute focus statement in the dispute focus statement set to obtain a plurality of similarities;

and taking the type corresponding to the sample dispute focus statement corresponding to the maximum value in the similarity as the type of the second target statement.

12. A method for generating a summary, comprising:

acquiring character information of a court trial to be detected;

determining at least one dispute focus based on elements of the plurality of element sets;

and outputting the summary information.

13. An apparatus for generating a summary, comprising:

the acquisition module is used for acquiring the character information of the court trial to be detected;

the determining module is used for determining an element set of each role in the court trial text information to be detected based on a legal knowledge element map to obtain a plurality of element sets, wherein elements in the element sets are used for reflecting the key points of the disputes of each role;

the analysis module is used for inputting the elements in the element sets into a dispute focus model for analysis to obtain at least one dispute focus;

the extracting module is used for extracting the summary information of the statement where the at least one dispute focus is located;

and the output module is used for outputting the summary information.

14. A non-volatile storage medium, comprising a stored program, wherein when the program runs, a device on which the storage medium is located is controlled to execute the summary generation method according to any one of claims 1 to 10.

15. A computer device, comprising:

a processor; and

a memory coupled to the processor for providing instructions to the processor for processing the following processing steps: acquiring character information of a court trial to be detected;

and outputting the summary information.