CN107562828B - Multi-source maritime information searching and conflict processing system and method - Google Patents
Multi-source maritime information searching and conflict processing system and method Download PDFInfo
- Publication number
- CN107562828B CN107562828B CN201710723562.2A CN201710723562A CN107562828B CN 107562828 B CN107562828 B CN 107562828B CN 201710723562 A CN201710723562 A CN 201710723562A CN 107562828 B CN107562828 B CN 107562828B
- Authority
- CN
- China
- Prior art keywords
- information
- module
- information data
- voice
- maritime
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The invention provides a multi-source maritime information searching and conflict processing system, which comprises a maritime information word bank module; the text searching module searches text information data about maritime affairs on the network; the voice input module receives voice information data to be recognized and preprocesses the voice information data; the voice recognition platform module recognizes the received voice information data based on the marine information keywords in the marine information word stock; the conflict processing module performs basic probability distribution on the text information data and the voice information data with conflicts by utilizing a DS evidence theory to obtain a reliability distribution function, and finally performs evidence combination to solve the conflicts of the text information data and the voice information data; the positioning customizing module positions the geographical position of the ship and selects information according to the geographical position; and a broadcasting module. The invention can customize marine information for the ship according to time and the area of the ship, and assists the driver to safely drive the ship.
Description
Technical Field
The invention belongs to the field of maritime information service, and particularly relates to a multi-source maritime information searching and conflict processing system and method.
Background
The rapid development of information technology, the appearance and rapid popularization of the internet mark that a global information society is gradually formed, and social economy and people's life increasingly depend on modern information technology. The establishment of the external release system of the marine traffic information is beneficial to sharing and utilizing the traffic dynamic and static information in the largest range and to the maximum extent by navigation ships, port and navigation enterprises, maritime management departments and the like, thereby realizing the safe optimized operation of the whole marine traffic system.
With the popularization and spread of networks, the information disclosure on the internet is recognized as a simple and efficient information disclosure form. The maritime management institutions at all levels also establish externally disclosed websites according to respective conditions, establish an online information disclosure column, and finish a carrier form necessary for online information disclosure by taking the construction of information disclosure catalogues, contents, online consultation, online forum and the like as marks.
In recent years, various maritime radio stations are established in coastal inland rivers of China, such as the "sound of maritime" column of the sun-uncovering radio station, the water safety information station of Yangtze river and the like. The broadcast receiving of various shipping safety information becomes the daily behavior of inland river ships in China, and the broadcast voice information also becomes one of the main ways for disclosing the maritime information.
At present, many researches on marine information services are conducted at home and abroad, but a personalized query mode cannot be conveniently provided for a user by a marine information publishing based on a VTS (vessel Traffic Operations Support system) or a marine information system based on a VTOSS (vessel Traffic Operations Support system), a ship driver cannot know marine information of a ship driving water area in real time, and serious consequences can be caused due to the loss of key marine information under the condition that people cannot avoid negligence.
It is known that weather information and channel information have their time-effects, and weather forecast for twenty-four hours and weather forecast for twelve hours in a certain area are likely to be different, and even more specifically, weather warning information may appear. And the information on the network and the information in the broadcast have different updating frequencies, so that the multi-source information has certain conflict, and how to process the conflict and obtain accurate maritime information is an urgent problem to be solved.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the system and the method for multi-source maritime information search and conflict processing are provided, and maritime information can be customized for a ship according to time and a region where the ship is located, so that a driver is assisted to safely drive the ship.
The technical scheme adopted by the invention for solving the technical problems is as follows: a multi-source maritime information searching and conflict processing system is characterized in that: it includes:
the marine information word bank module is used for storing key words in the marine field;
the text search module is used for searching text information data about maritime affairs on the network;
the voice input module is used for receiving voice information data to be recognized and preprocessing the voice information data;
the voice recognition platform module is used for recognizing the received voice information data based on the marine information keywords in the marine information word stock;
the conflict processing module is used for carrying out basic probability distribution on the text information data and the voice information data with conflicts by utilizing a DS evidence theory to obtain a reliability distribution function, and finally carrying out evidence combination to solve the conflicts of the text information data and the voice information data;
the positioning customization module is used for positioning the geographic position of the ship and selecting information according to the geographic position;
and the broadcasting module is used for broadcasting the information content after the conflict processing and the selection of the positioning module.
According to the system, the character searching module is specifically used for crawling by using a network crawler based on a TF-IDF algorithm so as to search character information data about maritime affairs on the network.
According to the system, the voice recognition platform module comprises a feature extraction module and a pattern matching recognition module; wherein the content of the first and second substances,
the feature extraction module is used for extracting information which is useful for recognition from the received voice information data, and taking the information into an acoustic model for matching to obtain pronunciation information of the voice information data; the acoustic model is trained by a maritime information word stock module through a hidden Markov model;
the pattern matching identification module is used for knowing pronunciation information of a hidden Markov model and voice information data to obtain a hidden state with the maximum probability contained in the pronunciation information, wherein the hidden state is identified characters; the known hidden Markov model is obtained by training the vocabulary contained in the maritime information word stock module.
According to the system, the positioning customizing module is specifically used for acquiring the position of the water area where the ship is located through the positioning system of the ship, selecting the marine information related to the located water area from the marine information which is continuously searched and processed before, and customizing a piece of marine information content unique to the current water area.
According to the system, the broadcasting module comprises a text broadcasting module and a voice broadcasting module.
A multi-source maritime information searching and conflict processing method is characterized by comprising the following steps: it comprises the following steps:
character searching: searching character information data about maritime affairs on the network;
and (3) voice input: receiving voice information data to be identified, and preprocessing the voice information data;
and (3) voice recognition: identifying the received voice information data based on the marine information keywords in the marine information word stock; the maritime information word bank module stores key words in the maritime field;
and (3) conflict processing: carrying out basic probability distribution on the text information data and the voice information data with conflicts by utilizing a DS evidence theory to obtain a reliability distribution function, and finally carrying out evidence combination to solve the conflicts of the text information data and the voice information data;
positioning and customizing: locating the geographical position of the ship and selecting information according to the geographical position;
broadcasting: and broadcasting the information content after the conflict processing and the selection of the positioning module.
According to the method, when characters are searched, the web crawler based on the TF-IDF algorithm is used for crawling.
According to the method, during voice recognition, information useful for recognition is extracted from received voice information data, and the information is taken into an acoustic model for matching to obtain pronunciation information of the voice information data; the acoustic model is trained by a maritime information word stock module through a hidden Markov model;
then, knowing pronunciation information of a hidden Markov model and voice information data, solving a hidden state with the maximum probability contained in the pronunciation information, wherein the hidden state is a recognized character; the known hidden Markov model is obtained by training the vocabulary contained in the maritime information word stock module.
According to the method, when positioning and customizing are carried out, the position of the water area where the ship is located is obtained through a positioning system of the ship, the marine information related to the located water area is selected from the marine information which is continuously searched and processed, and a piece of marine information content unique to the current water area is customized.
According to the method, when broadcasting, text broadcasting and voice broadcasting are respectively adopted.
The invention has the beneficial effects that: the method comprises the steps of acquiring various relevant marine information on a network and in broadcasting in real time, broadcasting the marine information after carrying out conflict processing based on an evidence theory and personalized customization and selection according to the position of a ship, and being capable of customizing the marine information for the ship according to time and the region where the ship is located and assisting a driver to drive the ship safely.
Drawings
FIG. 1 is a flowchart of a method according to an embodiment of the present invention.
FIG. 2 is a flow chart of a speech recognition process.
Fig. 3 is a schematic diagram of hidden markov model training acoustic models.
Detailed Description
The invention is further illustrated by the following specific examples and figures.
The invention provides a multi-source maritime information searching and conflict processing system, which comprises:
and the marine information word bank module is used for storing key words in the marine field.
And the text searching module is used for searching text information data about maritime affairs on the network. The word searching module is specifically used for crawling by using a web crawler based on a TF-IDF (Term Frequency-Inverse Document Term Frequency statistics) algorithm, so as to search word information data on the network about maritime affairs.
The TF-IDF improvement algorithm specifically comprises the following steps:
and taking words in the marine information word stock module as topic keywords, wherein the frequency of the topic keywords appearing in the document is TF frequency, and the distinguishing degree of the topic keywords in the document is IDF frequency. The calculation formula of the TF-IDF algorithm is as follows:
wherein: omegaijIs the weight of the ith word in the jth article; f. oftfIs the frequency of occurrence of the ith vocabulary in the jth document, called the vocabulary frequency; f. ofidfIs the frequency with which the topic appears in all documents, called the anti-lexical frequency. When f isidfThe larger the value, the more easily this vocabulary is distinguished from other documents in all documents, the higher the degree of recognition. Omega is obtained from the formula (1)ijThe vocabulary frequency needs to be increased if the weight of (2) is to be increased. For inverse vocabulary frequencies, the larger the value, the more concentrated the vocabulary is in some documents and is more easily distinguished from other documents.
In this embodiment, the focused web crawler using the TF-IDF algorithm will continuously download relevant marine information on the network and perform simple preprocessing according to the marine information thesaurus module as a theme.
And the voice input module is used for receiving voice information data needing to be recognized and preprocessing the voice information data.
And the voice recognition platform module is used for recognizing the received voice information data based on the marine information keywords in the marine information word stock.
The voice recognition platform module adopts a science and university news flying voice cloud open interface, the calling news flying voice cloud open interface is actually a service end for accessing an MSP (management service provider) platform, the service end provides services such as HTTP application, user management, voice service and the like, is positioned in a local area network, is externally and uniformly accessed to the Internet, and provides a unique access point for a client. Wherein: the HTTP server is responsible for sending a service request sent by the client to the business server, then the business server processes the service request according to a specific service type, an ISP voice application platform is called to obtain a specific voice service, then a processing result is returned to the HTTP server, and then the client is replied.
The MSP system mainly comprises four layers of a Speech application Interface (SPI), a Client (MSC), a Server (MSS) and a basic support (MSPINFrastrcture), and the four logic layers form a complete MSP system architecture from a user to a bottom layer of a Server operating system.
The voice recognition platform module comprises a feature extraction module and a mode matching recognition module; the feature extraction module is used for extracting information which is useful for recognition from the received voice information data, and taking the information into an acoustic model for matching to obtain pronunciation information of the voice information data; the acoustic model is trained by a maritime information word stock module through a hidden Markov model; the pattern matching identification module is used for knowing pronunciation information of a hidden Markov model and voice information data to obtain a hidden state with the maximum probability contained in the pronunciation information, wherein the hidden state is identified characters; the known hidden Markov model is obtained by training the vocabulary contained in the maritime information word stock module.
FIG. 3 is a schematic diagram of an acoustic model trained by a hidden Markov model, in this embodiment, a first module is used to train a hidden Markov-Gaussian mixture (HMM) -GMM model with training data, a modeling unit of the HMM-GMM model is a triplet state of speech features of the training data after being clustered by a phoneme decision tree, and the HMM-GMM model obtains a state transition probability of the triplet state by an Expectation Maximization (EM) algorithm; the second module is used for carrying out forced alignment on the training data speech features based on the HMM-GMM model to obtain the triple state information of the speech feature frame level; a third module for pre-training a deep neural network as the acoustic model to obtain parameters for initializing weights of hidden layers of the deep neural network; and the fourth module is used for training the deep neural network by adopting an error back propagation algorithm based on the speech feature frame-level state information of the speech features of the training data and updating the weight of each hidden layer.
Hidden Markov Models (HMM) are statistical models that are used to describe a Markov process with Hidden unknown parameters. The difficulty is to determine the implicit parameters of the process from the observable parameters. These parameters are then used for further analysis, such as pattern recognition. A hidden markov model is a type of markov chain whose states are not directly observable but observable through a sequence of observation vectors, each observation vector being represented as a variety of states by some probability density distribution, each observation vector being generated from a sequence of states having a corresponding probability density distribution. Thus, the hidden Markov model is a dual stochastic process- -a hidden Markov chain with a certain number of states and a set of display stochastic functions. HMMs have been used for speech recognition with great success since the 80's of the 20 th century. By the 90 s, HMMs were also introduced in computer text recognition and mobile communication core technologies "detection of multiple users". HMMs are also beginning to find applications in the fields of bioinformatics science, fault diagnosis, and the like.
Hidden Markov Models (HMMs) can be described by five elements, including 2 state sets and 3 probability matrices:
1. implicit State S
The Markov property is satisfied between the states, and the states are actually hidden states in the Markov model. These conditions are often not accessible by direct observation. (e.g., S1, S2, S3, etc.)
2. Observable state O
The association with implicit states in the model can be obtained by direct observation. (e.g., O1, O2, O3, etc., the number of observable states need not be consistent with the number of implied states.)
3. Initial state probability matrix pi
A probability matrix of the hidden state at the initial time t ═ 1 is shown, (for example, when t is 1, P (S1) ═ P1, P (S2) ═ P2, and P (S3) ═ P3, the initial state probability matrix pi ═ P1P2P 3).
4. Implicit state transition probability matrix A
Transition probabilities between states in HMM models are described. Where Aij ≦ P (Sj | Si),1 ≦ i, j ≦ n, indicates the probability that the state is Sj at time t +1, given that the state is Si at time t.
5. The state transition probability Matrix B (english name fusion Matrix, translated as a Confusion Matrix is less easily understood literally) is observed.
Let N represent the number of implicit states and M represent the number of observable states, then: bij ≦ P (Oi | Sj),1 ≦ i ≦ M,1 ≦ j ≦ N.
In general, a hidden markov model can be compactly represented by a (a, B, pi) triplet. Hidden markov models are in fact extensions of the standard markov models, adding a set of observable states and probabilistic relationships between these states and hidden states.
In this embodiment, training the acoustic model consists of the following steps: 1. and establishing an initial model. Specifically, training a hidden Markov-Gaussian mixture (HMM-GMM) model by using training data, wherein a modeling unit of the HMM-GMM model is a triplet state of speech features of the training data after being clustered by a phoneme decision tree, and the HMM-GMM model obtains the state transition probability of the triplet state by an expected maximum Expectation Maximization (EM) algorithm; 2. speech feature frame-level state information of speech features of training data is obtained. Specifically, based on the HMM-GMM model, the triplet states of the training data speech features are forcibly aligned, and the speech feature frame-level state information is obtained; preferably, the HMM-GMM model is based on the training data, and the triplet state of the speech feature is forcibly aligned to obtain the frame-level state information of the speech feature, specifically: and on the basis of the daily HMM-GMM model, corresponding the training data speech features to the most possible triple states thereof to obtain the speech feature frame-level state information. 3. And initializing each hidden layer weight of the deep neural network. Specifically, a deep neural network as the acoustic model is pre-trained to obtain parameters for initializing weights of hidden layers of the deep neural network; 4. and updating the weight of each hidden layer of the deep neural network. Specifically, the deep neural network is trained by adopting an error back propagation algorithm based on the triple state of the training data voice features, and the weight of each hidden layer is updated. Preferably, the pre-training of the deep neural network as the acoustic model to obtain the parameters for initializing the weights of the hidden layers of the deep neural network is specifically: and training layer by layer to be convergent by utilizing a restricted Boltzmann machine based on the training data, and initializing the weight of each hidden layer of the deep network by using the obtained parameters.
When the HMM-GMM model is used for voice recognition as an acoustic model, the posterior probability generated by the voice features through a deep neural network is converted into likelihood probability through a Bayes formula and is sent to a decoder for decoding, and a text sequence obtained after decoding is used as recognized speaking content. The effectiveness of speech recognition can be evaluated based on the difference between the content of the recognized utterance and the actual original speech. According to the effect, the performance of a deep neural network serving as an acoustic model in the speech recognition system can be evaluated, retraining can be considered when necessary, and even redesigning of the state transition probability in the HMM-GMM model can be considered.
And the conflict processing module is used for performing basic probability distribution on the text information data and the voice information data with conflicts by utilizing a DS evidence theory to obtain a confidence distribution function, and finally performing evidence combination to solve the conflicts of the text information data and the voice information data.
In this example, the evidence theory has the following implementation steps:
assuming that weather forecast information obtained from the network shows that the probability of raining in 6 hours in the future is 0.1 and the probability of not raining is 0.9; and the emergency weather warning information received by the broadcast shows that the probability of raining in the future 6 hours is 0.9, and the probability of not raining is 0.1.
Then, the probability of raining in the tomorrow, which is derived from evidence theory, is:
since the two evidences presented are diametrically opposite, the probability obtained is 0.5. In the embodiment, different evidences have different weights, and different results can be obtained according to the different weights, so that the result of the conflict processing can be obtained more accurately.
The weight value comes from the reliability of the information, the expert can be used for scoring in the initial stage, and the machine learning can be performed by data feedback in the later stage.
And the positioning customization module is used for positioning the geographic position of the ship and selecting the information. The positioning customizing module is specifically used for acquiring the position of the water area where the ship is located through a positioning system of the ship, selecting the marine information related to the located water area from the marine information which is continuously searched and processed before, and customizing a piece of marine information content unique to the current water area.
The broadcasting module is used for broadcasting the information content after conflict processing and selection by the positioning module and comprises a text broadcasting module and a voice broadcasting module.
A multi-source maritime information searching and conflict processing method is shown in FIG. 1, and comprises the following steps:
character searching: searching character information data about maritime affairs on the network; and when searching characters, crawling by using a web crawler based on the TF-IDF algorithm.
And (3) voice input: and receiving voice information data to be identified, and preprocessing the voice information data.
And (3) voice recognition: identifying the received voice information data based on the marine information keywords in the marine information word stock; the marine information word stock module stores key words in the marine field. During voice recognition, as shown in fig. 2, information useful for recognition is extracted from received voice information data, and the information is taken into an acoustic model for matching to obtain pronunciation information of the voice information data; the acoustic model is trained by a maritime information word stock module through a hidden Markov model; then, knowing pronunciation information of a hidden Markov model and voice information data, solving a hidden state with the maximum probability contained in the pronunciation information, wherein the hidden state is a recognized character; the known hidden Markov model is obtained by training the vocabulary contained in the maritime information word stock module.
And (3) conflict processing: and performing basic probability distribution on the text information data and the voice information data with conflicts by utilizing a DS evidence theory to obtain a reliability distribution function, and finally performing evidence combination to solve the conflicts of the text information data and the voice information data.
Positioning and customizing: the geographical location of the vessel is located and the information is picked up accordingly. When positioning and customizing are carried out, the position of the water area where the ship is located is obtained through a positioning system of the ship, the marine information related to the located water area is selected from the marine information which is continuously searched and processed in the past, and unique marine information content of the current water area is customized.
Broadcasting: and broadcasting the information content after the conflict processing and the selection of the positioning module. When broadcasting, respectively adopting character broadcasting and voice broadcasting.
The invention can automatically acquire the network character maritime information and the broadcast voice maritime information of the water area where the ship runs in real time, obtain the most accurate result after conflict processing, and report various maritime information of the water area where the ship passes to a driver in real time in a character and voice mode after personalized customization based on the position information, thereby providing assistance for safe driving of the ship.
The above embodiments are only used for illustrating the design idea and features of the present invention, and the purpose of the present invention is to enable those skilled in the art to understand the content of the present invention and implement the present invention accordingly, and the protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes and modifications made in accordance with the principles and concepts disclosed herein are intended to be included within the scope of the present invention.
Claims (10)
1. A multi-source maritime information searching and conflict processing system is characterized in that: it includes:
the marine information word bank module is used for storing key words in the marine field;
the text search module is used for searching text information data about maritime affairs on the network;
the voice input module is used for receiving voice information data to be recognized and preprocessing the voice information data;
the voice recognition platform module is used for recognizing the received preprocessed voice information data based on the marine information keywords in the marine information word stock;
the conflict processing module is used for carrying out basic probability distribution on the character information data with conflict and the recognized voice information data by utilizing a DS evidence theory to obtain a reliability distribution function, and finally carrying out evidence combination to solve the conflict between the character information data and the recognized voice information data;
the positioning customization module is used for positioning the geographic position of the ship and selecting information according to the geographic position;
and the broadcasting module is used for broadcasting the information content after the conflict processing and the selection of the positioning module.
2. The multi-source maritime information search and conflict handling system of claim 1, wherein: the character searching module is specifically used for crawling by using a web crawler based on a TF-IDF algorithm so as to search character information data about maritime affairs on the network.
3. The multi-source maritime information search and conflict handling system of claim 1, wherein: the voice recognition platform module comprises a feature extraction module and a pattern matching recognition module; wherein the content of the first and second substances,
the feature extraction module is used for extracting information which is useful for recognition from the received voice information data, and taking the information into an acoustic model for matching to obtain pronunciation information of the voice information data; the acoustic model is trained by a maritime information word stock module through a hidden Markov model;
the pattern matching identification module is used for knowing pronunciation information of a hidden Markov model and voice information data to obtain a hidden state with the maximum probability contained in the pronunciation information, wherein the hidden state is identified characters; the known hidden Markov model is obtained by training the vocabulary contained in the maritime information word stock module.
4. The multi-source maritime information search and conflict handling system of claim 1, wherein: the positioning customization module is specifically used for acquiring the position of a water area where a ship is located through a positioning system of the ship, selecting the marine information related to the located water area from the marine information which is continuously searched and processed before, and customizing a piece of marine information content unique to the current water area.
5. The multi-source maritime information search and conflict handling system of claim 1, wherein: the broadcasting module comprises a text broadcasting module and a voice broadcasting module.
6. A multi-source maritime information searching and conflict processing method is characterized by comprising the following steps: it comprises the following steps:
character searching: searching character information data about maritime affairs on the network;
and (3) voice input: receiving voice information data to be identified, and preprocessing the voice information data;
and (3) voice recognition: based on the marine information keywords in the marine information word stock, identifying the received preprocessed voice information data; the maritime information word bank module stores key words in the maritime field;
and (3) conflict processing: carrying out basic probability distribution on the character information data with conflict and the recognized voice information data by utilizing a DS evidence theory to obtain a reliability distribution function, and finally carrying out evidence combination to solve the conflict between the character information data and the recognized voice information data;
positioning and customizing: locating the geographical position of the ship and selecting information according to the geographical position;
broadcasting: and broadcasting the information content after the conflict processing and the selection of the positioning module.
7. The multi-source maritime information search and conflict handling method of claim 6, wherein: and when searching characters, crawling by using a web crawler based on the TF-IDF algorithm.
8. The multi-source maritime information search and conflict handling method of claim 6, wherein: during voice recognition, information useful for recognition is extracted from received voice information data, and the information is taken into an acoustic model for matching to obtain pronunciation information of the voice information data; the acoustic model is trained by a maritime information word stock module through a hidden Markov model;
then, knowing pronunciation information of a hidden Markov model and voice information data, solving a hidden state with the maximum probability contained in the pronunciation information, wherein the hidden state is a recognized character; the known hidden Markov model is obtained by training the vocabulary contained in the maritime information word stock module.
9. The multi-source maritime information search and conflict handling method of claim 6, wherein: when positioning and customizing are carried out, the position of the water area where the ship is located is obtained through a positioning system of the ship, the marine information related to the located water area is selected from the marine information which is continuously searched and processed in the past, and unique marine information content of the current water area is customized.
10. The multi-source maritime information search and conflict handling method of claim 6, wherein: when broadcasting, respectively adopting character broadcasting and voice broadcasting.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710723562.2A CN107562828B (en) | 2017-08-22 | 2017-08-22 | Multi-source maritime information searching and conflict processing system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710723562.2A CN107562828B (en) | 2017-08-22 | 2017-08-22 | Multi-source maritime information searching and conflict processing system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107562828A CN107562828A (en) | 2018-01-09 |
| CN107562828B true CN107562828B (en) | 2020-10-30 |
Family
ID=60976239
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710723562.2A Active CN107562828B (en) | 2017-08-22 | 2017-08-22 | Multi-source maritime information searching and conflict processing system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107562828B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108897778B (en) * | 2018-06-04 | 2021-12-31 | 创意信息技术股份有限公司 | Image annotation method based on multi-source big data analysis |
| CN110110038B (en) * | 2018-08-17 | 2023-04-18 | 平安科技（深圳）有限公司 | Telephone traffic prediction method, device, server and storage medium |
| CN110047330A (en) * | 2019-04-16 | 2019-07-23 | 湖北地信科技集团股份有限公司 | Processing method, user equipment, storage medium and the device of navigation data |
| CN110600007B (en) * | 2019-08-30 | 2022-02-18 | 中华人民共和国天津海事局 | Ship recognition and positioning system and method based on voice |
| CN112086090A (en) * | 2020-09-14 | 2020-12-15 | 大连海事大学 | Voice recognition method of ship electromechanical virtual simulation system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1185975A1 (en) * | 1999-04-20 | 2002-03-13 | Verance Corporation | Broadcast speech recognition system for keyword monitoring |
| US20070018953A1 (en) * | 2004-03-03 | 2007-01-25 | The Boeing Company | System, method, and computer program product for anticipatory hypothesis-driven text retrieval and argumentation tools for strategic decision support |
| CN101996157A (en) * | 2010-10-23 | 2011-03-30 | 山东科技大学 | Multisource information fusion method in evidence high-conflict environment |
| US20120005148A1 (en) * | 2010-06-30 | 2012-01-05 | Microsoft Corporation | Integrating specialized knowledge sources into a general search service |
| CN103730016A (en) * | 2013-12-17 | 2014-04-16 | 深圳先进技术研究院 | Traffic information publishing system and method |
| CN106354860A (en) * | 2016-09-06 | 2017-01-25 | 中国传媒大学 | Method for automatically labelling and pushing information resource based on label sets |
-
2017
- 2017-08-22 CN CN201710723562.2A patent/CN107562828B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1185975A1 (en) * | 1999-04-20 | 2002-03-13 | Verance Corporation | Broadcast speech recognition system for keyword monitoring |
| US20070018953A1 (en) * | 2004-03-03 | 2007-01-25 | The Boeing Company | System, method, and computer program product for anticipatory hypothesis-driven text retrieval and argumentation tools for strategic decision support |
| US20120005148A1 (en) * | 2010-06-30 | 2012-01-05 | Microsoft Corporation | Integrating specialized knowledge sources into a general search service |
| CN101996157A (en) * | 2010-10-23 | 2011-03-30 | 山东科技大学 | Multisource information fusion method in evidence high-conflict environment |
| CN103730016A (en) * | 2013-12-17 | 2014-04-16 | 深圳先进技术研究院 | Traffic information publishing system and method |
| CN106354860A (en) * | 2016-09-06 | 2017-01-25 | 中国传媒大学 | Method for automatically labelling and pushing information resource based on label sets |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107562828A (en) | 2018-01-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107562828B (en) | Multi-source maritime information searching and conflict processing system and method | |
| CN107846350B (en) | Method, computer readable medium and system for context-aware network chat | |
| Kim et al. | Two-stage multi-intent detection for spoken language understanding | |
| WO2022057712A1 (en) | Electronic device and semantic parsing method therefor, medium, and human-machine dialog system | |
| CN111241237B (en) | Intelligent question-answer data processing method and device based on operation and maintenance service | |
| CN113906433A (en) | Clarifying questions for rewriting ambiguous user utterances | |
| CN112800170A (en) | Question matching method and device and question reply method and device | |
| CN108763510A (en) | Intension recognizing method, device, equipment and storage medium | |
| CN110457689B (en) | Semantic processing method and related device | |
| CN109887484A (en) | A kind of speech recognition based on paired-associate learning and phoneme synthesizing method and device | |
| Harwath et al. | Zero resource spoken audio corpus analysis | |
| CN114676255A (en) | Text processing method, device, equipment, storage medium and computer program product | |
| CN113326702B (en) | Semantic recognition method, semantic recognition device, electronic equipment and storage medium | |
| CN110085217A (en) | Phonetic navigation method, device and terminal device | |
| CN113449084A (en) | Relationship extraction method based on graph convolution | |
| Lan et al. | Semi-supervised training using adversarial multi-task learning for spoken language understanding | |
| CN116018791A (en) | Multi-person call using single request in assistant system | |
| CN114756675A (en) | Text classification method, related equipment and readable storage medium | |
| CN113569011B (en) | Training method, device and equipment of text matching model and storage medium | |
| CN110246494A (en) | Service request method, device and computer equipment based on speech recognition | |
| CN114091454A (en) | Method for extracting place name information and positioning space in internet text | |
| CN116522905B (en) | Text error correction method, apparatus, device, readable storage medium, and program product | |
| CN116450799A (en) | Intelligent dialogue method and equipment applied to traffic management service | |
| Agarwal et al. | Lidsnet: A lightweight on-device intent detection model using deep siamese network | |
| CN116384405A (en) | Text processing method, text classification method and emotion recognition method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |