WO2021031383A1 - Procédé et appareil de jugement auxiliaire intelligent, et dispositif informatique et support de stockage - Google Patents

Procédé et appareil de jugement auxiliaire intelligent, et dispositif informatique et support de stockage Download PDF

Info

Publication number
WO2021031383A1
WO2021031383A1 PCT/CN2019/116487 CN2019116487W WO2021031383A1 WO 2021031383 A1 WO2021031383 A1 WO 2021031383A1 CN 2019116487 W CN2019116487 W CN 2019116487W WO 2021031383 A1 WO2021031383 A1 WO 2021031383A1
Authority
WO
WIPO (PCT)
Prior art keywords
text
trial
original
target
data
Prior art date
Application number
PCT/CN2019/116487
Other languages
English (en)
Chinese (zh)
Inventor
胡文成
赵付利
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2021031383A1 publication Critical patent/WO2021031383A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/18—Legal services
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

  • This application relates to the field of artificial intelligence technology, in particular to an intelligent auxiliary trial method, device, computer equipment and storage medium.
  • the inventor proposes an intelligent auxiliary trial system that can help to reduce the workload of judicial staff (including clerks and judges) in the trial process of cases and is applied to the intelligent auxiliary trial system Intelligent auxiliary trial method.
  • the embodiments of the present application provide an intelligent auxiliary trial method, device, computer equipment, and storage medium to solve the current problem of large workload of judicial staff in the judgment process.
  • An intelligent auxiliary trial method including:
  • An intelligent auxiliary trial device including:
  • the case description information acquisition module is used to obtain the court trial transcript file corresponding to the case to be judged from the database, and extract the case description information from the court trial transcript file;
  • the target word segmentation result acquisition module is configured to use a word segmentation tool to segment the case description information to obtain a target word segmentation result, where the target word segmentation result includes multiple target word segmentation;
  • the target keyword determining module is configured to query the keyword database based on each target word segmentation, and determine the original keywords stored in the keyword database that match the target word segmentation as target keywords;
  • the target case type determination module is configured to query the case type information database according to the target keyword, and obtain at least one target case type that matches the target keyword;
  • the target push law article acquisition module is used to acquire prior knowledge matching at least one of the target case types from the prior knowledge base, and use the intelligent law article push model to process the prior knowledge to obtain the target push method Article;
  • the recommended trial viewpoint acquisition module is configured to use a semantic analysis model to perform semantic analysis on the case description information, obtain standardized information, query the trial viewpoint database based on the standardized information, and obtain corresponding recommended trial viewpoints;
  • the trial suggestion file acquisition module is used to push the law and the recommended trial viewpoint according to the target, and acquire and display the trial suggestion file.
  • a computer device includes a memory, a processor, and computer-readable instructions stored in the memory and capable of running on the processor, and the processor implements the following steps when the processor executes the computer-readable instructions:
  • One or more non-volatile readable storage media storing computer readable instructions, the computer readable storage medium storing computer readable instructions, and when the computer readable instructions are executed by one or more processors, Make the one or more processors execute the following steps:
  • Fig. 1 is a schematic diagram of an application environment of an intelligent assisted trial method in an embodiment of the present application
  • Fig. 2 is a flow chart of the intelligent assisted trial method in an embodiment of the present application
  • FIG. 3 is another flowchart of the intelligent assisted trial method in an embodiment of the present application.
  • FIG. 4 is another flowchart of the intelligent assisted trial method in an embodiment of the present application.
  • Fig. 5 is a functional block diagram of the intelligent auxiliary trial device in an embodiment of the present application.
  • Fig. 6 is a schematic diagram of a computer device in an embodiment of the present application.
  • the intelligent auxiliary trial method provided by the embodiment of the present application can be applied to the application environment shown in FIG. 1.
  • the intelligent assisted trial method is applied in an intelligent assisted trial system.
  • the intelligent assisted trial system includes a client, a microphone, and a server as shown in FIG. 1. Both the client and the microphone communicate with the server through a network, where the client
  • the client is also called the client, which refers to the program that corresponds to the server and provides local services to the client.
  • the client can be installed on, but not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices.
  • the client is a terminal used to implement human-computer interaction with the parties in the court trial, and the microphone is used to collect voice data of the parties in the court trial.
  • clients and microphones are installed on the judge’s bench, the plaintiff’s seat, the court’s seat, and the third party’s seat, while the clerk’s seat has a client.
  • the server can be implemented as an independent server or a server cluster composed of multiple servers.
  • the client and the server in the intelligent auxiliary trial system communicate through the network to realize the processing of the court trial transcripts to obtain the result of the judgment proposal, so that the judge can make the judgment based on the result of the judgment proposal It is recommended that the results be judged accordingly, so as to reduce the workload of judges and improve their work efficiency.
  • an intelligent assisted trial method is provided. The method is applied to the server in Figure 1 as an example for description, including the following steps:
  • S201 Obtain court trial transcript files corresponding to the case to be judged from the database, and extract case description information from the court trial transcript files.
  • the pending cases refer to cases that need to be tried and handled.
  • the court hearing transcript is a document that records the entire litigation activities carried out by all the parties to the court hearing during the court hearing.
  • the court trial transcript file in this embodiment is the court trial transcript file formed in the foregoing embodiment and stored in the database.
  • Case description information refers to the description information extracted from the court trial transcripts that is related to the pending case and will affect the trial result.
  • This court trial transcript is a document that collects information such as facts and evidence of the case to be judged by the parties in the court trial based on the standard court trial transcript template during the court trial.
  • the standard court trial transcript template is a pre-set transcript template for recording case information corresponding to each process in the court trial process.
  • the trial transcript file is a transcript file corresponding to the pending case collected by the courts at all levels according to the standard court trial transcript template
  • the court trial transcript file has a standard format, which can help quickly extract case description information.
  • the transcript of a court hearing records the following content: the facts that the two parties have no dispute are..., the focus of the dispute between the two parties is...; in the process of obtaining case description information from the court hearing transcript, a text matching algorithm can be used to match the original
  • the courts have undisputed facts" and "the focus of disputes between the two parties have” fields.
  • the content after these fields are determined as the facts ascertained by the court and the focus of the dispute between the parties.
  • These content are the descriptive information that affects the outcome of the trial. It serves as case description information.
  • the case description information can also be filled into the corresponding position of the judgment document template, and the final judgment recommendation result determined according to this embodiment is obtained.
  • the case basis that is, the specific applicable legal provisions
  • the case basis and the judgment point of view are filled into the corresponding position of the judgment document template to quickly obtain the corresponding target judgment document.
  • the target judgment document is the final judgment document corresponding to the pending case, such as a civil judgment and a criminal judgment.
  • S202 Use a word segmentation tool to segment the case description information to obtain a target segmentation result, and the target segmentation result includes multiple target segmentation.
  • the word segmentation tool is a tool used to implement Chinese word segmentation of the text.
  • Chinese Word Segmentation refers to the segmentation of a sequence of Chinese characters into multiple individual words. Word segmentation refers to the process of recombining consecutive word sequences into word sequences according to certain specifications.
  • the target segmentation result refers to the result of multiple segmentation formed by the case description information.
  • the target word segmentation refers to the word segmentation finally formed according to the case description information.
  • the word segmentation tool may be an open source tool commonly used in the market for word segmentation of Chinese text-stuttering word segmentation.
  • Stuttering Chinese word segmentation is used to segment the case description information.
  • the segmentation process can support three word segmentation modes including: (1) Accurate mode: Try to cut the sentence most accurately, suitable for text analysis, but low efficiency. (2) Full mode: Scan all the words in the sentence that can be formed into words. The word segmentation speed is fast, but it cannot solve the ambiguity problem.
  • step S202 specifically includes the following steps:
  • S2011 Use the search engine mode of the stuttering word segmentation tool to perform text segmentation on the case description information, and obtain the text segmentation result.
  • the text segmentation result includes N first-level word segmentation.
  • the server uses the search engine mode of the stammering word segmentation tool to perform text segmentation of the case description information to quickly obtain the text segmentation result.
  • the text segmentation result can be understood as the result of the word segmentation using the conventional search engine mode, which is before optimization Word segmentation result.
  • the first-level word segmentation is the word segmentation that constitutes the text segmentation result, and N is the number of first-level word segmentation in the text word segmentation result.
  • the obtained text segmentation results include the following 5 first-level word segmentation: 1. Defendant, 2. .None, 3. Behavior, 4. Ability, 5. Behavior Ability. From the above text segmentation results, it can be seen that in three consecutive first-level participles (from the third first-level participle to the fifth first-level participle), the third first-level participle "behavior" and the fourth first-level participle "ability" After the combination, it repeats the meaning of the fifth first-level word segmentation "behavior", and the word segmentation result is inaccurate. If the subsequent semantic analysis is directly based on the text segmentation result, the efficiency and accuracy of the subsequent analysis may be affected.
  • Superposition means that in two adjacent first-level participles, at least one Chinese character at the end of the previous first-level participle overlaps with at least one Chinese character at the front end of the next first-level participle, that is, two adjacent first-level participles.
  • Grade participles can be spliced based on overlapping Chinese characters, leaving only one overlapping character or word to form a concatenated word, for example: the three consecutive first-level participles are: “Management”, “Technology” and “Work” in " The overlapping Chinese character in the two first-level participles of "Management” and “Technology” is " ⁇ ", and the overlapping Chinese character of "Technology” and “ ⁇ ” is “ ⁇ ”.
  • the three first-level participles “management”, “science and engineering” and “work” can be superimposed to form a new word: "management work”.
  • “Combine” refers to the direct merging of two first-level participles without removing the merged form of the participles of repeated Chinese characters.
  • the two first-level participles are "management” and "work”, and the compound word “management work” can be obtained by merging the aforementioned two first-level participles.
  • the case description information is: the court masters trade secrets during the management work; the search engine mode of the stuttering word segmentation tool is used to segment the case description information, and the text segmentation results obtained are: 1. The court, 2. In, 3. Management, 4. Science and engineering, 5. Work, 6. Management work, 7. Process, 8. In, 9. Master, 10. Business, 11. Secret, 12. Trade secret.
  • the superposition of "management", “science and engineering” and “work” equals “management work”
  • the combination of "management” and “work” equals “management work”. Therefore, only “management” and “work” are retained.
  • Two first-level participles are used as optimized target participles, and the two first-level participles of "science and engineering” and “management work” are deleted.
  • the case description information is: this clause is only for candidates with industry experience; after text segmentation is performed on the case description information using the search engine mode of the stuttering word segmentation tool, the text segmentation results obtained are: 1. This, 2. . Clause, 3. Only, 4, Targeting, 5. Yes, 6, Same industry, 7. Experience, 8. Same industry experience, 9, of, 10, Candidate, 11, Pick, 11, Candidate. Among them, the superposition of "candidate” and “chosen” is equal to “candidate”, and there is no situation that the combination is equal to “candidate”. At this time, delete "candidate” and “choose person” and only keep the “candidate” The first level participle.
  • the case description information is: the court masters trade secrets during the management work; the search engine mode of the stuttering word segmentation tool is used to segment the case description information, and the text segmentation results obtained are: 1. The court, 2. In, 3. Management, 4. Science and Engineering, 5. Work, 6. Management Work, 7. Process, 8. In, 9. Master, 10. Business, 11. Secret, 12. Trade Secret.
  • the combination of "commercial” and “secret” is equal to "commercial secret", and there is no overlap. Therefore, only the two first-level participles of "commercial” and “secret” are retained as the optimized target participles, and " "Trade secret” is a first-level participle.
  • first-level segmentation words with semantic repetition are analyzed to determine whether there is a combination or superposition of consecutive first-level segmentation words that can constitute a continuous
  • the long words immediately following the first-level participles of, and the first-level participles or the final long words are processed according to the different combinations or superpositions of the continuous first-level participles to obtain more accurate word segmentation results, reducing repeated words or uselessness Word, in order to realize the optimization of the word segmentation results under the premise of ensuring the efficiency of word segmentation, and improve the accuracy of word segmentation.
  • S203 Query the keyword database based on each target word segmentation, and determine the original keyword stored in the keyword database that matches the target word segmentation as the target keyword.
  • the keyword library is a preset database for storing original keywords.
  • the original keywords are pre-set keywords that may affect the identification of the case type.
  • the corresponding relationship between the original keyword and at least one synonym is stored in the keyword library.
  • the server uses a string matching algorithm or other matching algorithms to query whether the target word segment exists in the keyword database (it can be the original keyword or its corresponding synonym), and if the target word segment exists, it The original keyword corresponding to the target word segmentation is determined as the target keyword. For example, there are synonym groups A1, A2, and A3 in the original keyword database, and A1 is the original keyword. If the target word segmentation is the same as any of A1, A2, and A3, then A1 is determined as the target word segmentation corresponding The target keywords.
  • step S203 specifically includes the following steps:
  • S2031 Query the thesaurus according to the target word segmentation, and obtain the target synonyms corresponding to the target word segmentation.
  • the thesaurus is used to store a pre-set database for storing synonym relationships.
  • the target synonyms are the synonyms recorded in the thesaurus that have a synonym relationship with the target segmentation.
  • a synonym group with a synonym relationship is pre-stored in the thesaurus, and the server can use a string matching algorithm to match the target word segmentation, and then obtain a synonym having a synonym relationship with the target word segmentation and determine it as the target synonym.
  • S2032 Query the keyword database according to the target word segmentation and target synonyms, determine whether there is an original keyword matching the target word segmentation or the target synonym, and if the original keyword exists, determine the original keyword as the target keyword.
  • the criminal Law stipulates that “for the purpose of committing a crime, tools and conditions are prepared for crime. For prepared criminals, they can be given a lighter, mitigated punishment or exempt from punishment in the same way as the completed offender.”
  • the keywords in the preparation tools and manufacturing conditions are whether to determine whether It is the key word for the situation of "criminal preparation", and during the court hearing, the parties in the court hearing can use other synonyms to express this meaning. Therefore, the server needs to query the keyword database based on the target word segmentation and target synonyms to determine whether there is an original keyword matching the target word segmentation or target synonym in the keyword database. If the original keyword exists, it will be determined as the target keyword. In order to expand the search scope of the target word segmentation, as many target keywords as possible are determined from the case description information to improve the accuracy of the subsequent judgment and suggestion results obtained.
  • S204 Query the case type information database according to the target keyword, and obtain at least one target case type that matches the target keyword.
  • the case type information database is an information database used to store different case types and their corresponding keywords. Since different case types (such as different types of crimes in the criminal law) may correspond to the same keyword during the trial process, each keyword and its corresponding case type will be stored in the form of key-value pairs in the case type information database , So that after obtaining the target keyword, the server can query the case type information table according to the target keyword to determine all case types containing the target keyword as the target case type.
  • S205 Obtain prior knowledge that matches at least one target case type from the prior knowledge base, and obtain the target push rule based on the prior knowledge.
  • the priori knowledge base is a database constructed based on priori knowledge extracted from historical judgment data, and each historical judgment data is specific to a specific case type.
  • the server queries the prior knowledge base according to at least one case type, and queries the prior knowledge matching the at least one target case type from the prior knowledge base, thereby performing prior knowledge screening.
  • the prior knowledge includes case description information in historical judgment data corresponding to at least one target case type and corresponding case judgment results.
  • the a priori knowledge base stores all case knowledge with too much content, if all the data in the prior knowledge base is vectorized for each case, there may be a problem of low execution efficiency. Therefore, it can be determined by the target keywords At least one target case type, and then use at least one target case type to filter the prior knowledge in the prior knowledge base to determine its corresponding prior knowledge, so as to perform the subsequent steps S206 and S207 to avoid the prior knowledge All prior knowledge in the library is processed to improve execution efficiency.
  • the target push law specifically refers to extracting from all prior knowledge the judgment basis used in the judgment result of the case (ie the specific application of the law), and the application of all judgment basis The number of times is counted and sorted, and finally a target push rule based on the number of applications from more to less is formed, so that the judge can understand the basis of the judgment of relevant historical judgment data, so as to save the time of consulting relevant materials in the judgment process and reduce the workload.
  • S206 Use the semantic analysis model to perform semantic analysis on the case description information, obtain standardized information, and query the judgment opinion database corresponding to the target case type based on the standardized information, and obtain corresponding recommended judgment opinions.
  • the semantic analysis model is a model constructed based on the NLP (Natural Language Processing) technology for analyzing the semantics of text.
  • Standardized information refers to the information extracted from the case description information that may affect the trial of the case.
  • the judgment viewpoint database is a database used to record the historical description information extracted from the historical judgment data and the association relationship between the corresponding historical judgment viewpoints.
  • the server uses a semantic analysis model to perform semantic analysis on case description information to extract standardized information therein.
  • a semantic analysis model to describe the case.
  • information that matches the age of the court will be screened out as standardized information.
  • the similarity is sorted according to the text similarity, and the top N (the number of N can be set independently) historical judgment viewpoints corresponding to the historical description information with the greater text similarity are used as recommended judgment viewpoints so that the judge can judge In the course of the case, timely understand the historical trial viewpoints of similar cases, and combine the actual situation of the pending case to make a judgment in a timely manner, reducing the workload of the judge to consult relevant materials. Understandably, querying the trial viewpoint database corresponding to the target case type according to the standardized information can effectively reduce the amount of data comparison between the standardized information and the historical description information in the trial viewpoint database, and help improve the efficiency of data processing.
  • the server pushes the acquired at least one target law and the acquired at least one recommended trial viewpoint as a trial suggestion file corresponding to the case to be judged.
  • the trial suggestion file is a case description extracted by the system through the court trial transcript.
  • the information is analyzed, and the targets determined in the historical judgment data related to the description of the case are pushed to the legal articles and recommended trial viewpoints, so as to provide a reference for the judge to judge, reduce the judge's process of consulting related cases and reduce their workload.
  • the case description information after the case description information is quickly extracted from the court hearing transcript, the case description information can be segmented and keyword matching processing can quickly obtain its corresponding target keywords, and use the The target keyword determines the corresponding at least one target case type, so that prior knowledge matching at least one target case type can be selected from the prior knowledge base, and the corresponding target push law can be quickly obtained based on the prior knowledge , To speed up the efficiency of obtaining the judgment proposal documents.
  • the intelligent auxiliary trial method before step S201, that is, before obtaining the court trial transcript file corresponding to the case to be judged from the database, the intelligent auxiliary trial method further includes:
  • the data update task includes the original case type and task update time.
  • the data update task is a task used to update the prior knowledge base and target judgment model.
  • the original case type refers to the case type targeted by the data update task.
  • the task update time refers to the time when the data update task was executed last time.
  • the judgment basis corresponding to the target case type specifically refers to the legal basis corresponding to the target case type, including legal provisions, regulations, and judicial interpretations.
  • the judgment basis corresponding to the original case type is changed after the task update time, it means that the judgment basis corresponding to the original case type has changed after the task update time, and the subsequent case trial process New laws, regulations, and judicial interpretations will be applied to the judgment basis. Therefore, the prior knowledge base and target judgment model need to be updated.
  • the execution time of the judgment basis is determined as the time of change, and the time of change and system are obtained
  • the historical judgment data corresponding to the original case type between the current time and the data to be processed are determined based on the historical judgment data, which helps to improve the timeliness and accuracy of the target judgment model trained subsequently.
  • the to-be-processed data is specifically training data used to update the a priori knowledge base and the trial opinion base.
  • the preset period is a preset period for collecting data, which can be set to three months, half a year, or one year.
  • the judgment basis corresponding to the original case type does not change after the task update time, it means that the old trial basis is still applied in the subsequent case trial process, but as time changes, the application of these trial basis
  • the standard may change, such as the amount of compensation.
  • the historical judgment data corresponding to the original case type in the preset period before the current time of the system is used to determine the data to be processed to ensure the timeliness of the data to be processed. This helps to improve the timeliness of the subsequent training of the target judgment model.
  • the data to be processed includes a target area, which can be understood as the administrative area where the trial court corresponding to the historical judgment data is located, such as Guangdong province or Shenzhen.
  • a target area which can be understood as the administrative area where the trial court corresponding to the historical judgment data is located, such as Guangdong province or Shenzhen.
  • the historical judgment data corresponding to the same target area and the original case type is determined as the data to be processed.
  • the preset number threshold is a preset number threshold. If the target data volume of the to-be-processed data corresponding to any target area is greater than the preset number threshold, it means that there are more cases in the target area that are tried by courts at all levels corresponding to the original case type. Therefore, the same target area
  • the historical judgment data corresponding to the original case type is determined as the data to be processed, so that a priori knowledge base corresponding to the target area and the original case type is constructed based on the data to be processed, and a priori knowledge base corresponding to the target area and the original case type is constructed. In order to make the a priori knowledge base and the trial viewpoint database have greater reference significance and more pertinence for the pending cases corresponding to the target case type in the target area.
  • the historical judgment data corresponding to the same original case type is determined as the data to be processed.
  • the target data volume of the to-be-processed data corresponding to any target area is not greater than the preset number threshold, it means that the number of cases corresponding to the original case type in the courts at all levels in the target area is small, and the same original case type is corresponding
  • the historical judgment data is determined as the data to be processed, the a priori knowledge base corresponding to the original case type is constructed, and the trial viewpoint database corresponding to the original case type is trained, so that the The data to be processed is not limited to the same target area, but is only divided according to the original case type.
  • S304 Extract prior information from the data to be processed, and construct a priori knowledge base corresponding to the target case type based on the prior information.
  • the data to be processed is historical judgment data corresponding to the original case type.
  • the server extracts a priori information from the data to be processed. Specifically, it can be understood as extracting the description of the case and the result of the case from the data to be processed, and then extracting keywords from the description of the case and the result of the case, using keys specifically
  • the key-value form defines these keywords; then, the extracted keywords are used to construct a priori knowledge base corresponding to the original case type, specifically the key-value pairs extracted from all the data to be processed Stored in the database to construct a priori knowledge base corresponding to the original case type.
  • Key can specifically be a keyword in the case description that affects the judgment result, and Value is the judgment result in the case judgment result.
  • S305 Extract historical description information and historical judgment viewpoints from the data to be processed, and construct a judgment viewpoint database corresponding to the original case type based on the historical description information and historical judgment viewpoints.
  • the historical description information is the information extracted from the data to be processed that may affect the trial of the case. Specifically, it is extracted from the case description information of the data to be processed using a semantic analysis model.
  • the historical trial viewpoint is to extract the judge's trial viewpoint of the historical case from the data to be processed.
  • the server uses the historical description information and historical judgment opinions as a set of training data and inputs them to common CNN (Convolutional Neural Network) or RNN (Recurrent Neural Network) Carry out model training to update the model parameters, so as to obtain the trial opinion library corresponding to the original case type.
  • CNN Convolutional Neural Network
  • RNN Recurrent Neural Network
  • the intelligent auxiliary trial method according to whether the judgment basis corresponding to the original case type has changed after the task update time, the historical judgment data of different time periods is collected and determined to be used to construct prior knowledge To ensure the timeliness of the data to be processed, the a priori knowledge base and the trial view database are updated to ensure that the target push method is obtained separately based on the prior knowledge base and the trial view database. The timeliness and accuracy of the article and recommended trial opinions.
  • the intelligent assisted trial method further includes the following steps:
  • S401 Display the trial prompt text corresponding to the current trial link in the standard court trial transcript template on the client, and obtain the original voice data collected by the microphone and corresponding to the trial prompt text.
  • the standard court trial transcript template generally includes pre-trial preparation, announcement of the opening, court investigation and court debate, etc.
  • the court investigation stage it specifically includes the statement of the parties, the summary of the focus of the dispute, the cross-examination of evidence around the focus, the witness testimony in court, the reading of expert opinions and Corresponding instructions are provided for each link of the inspection transcript and application for appraisal.
  • This guiding speech is generally the speech corresponding to the guiding question that the judge guides other parties (such as the plaintiff, lawyer or witness) to respond, such as "the plaintiff first presents to the court around your litigation request".
  • Trial prompt text refers to the guiding words corresponding to the current trial link in the standard court trial transcript template.
  • the current trial link refers to the ongoing link in the court hearing the case, such as the statement of the parties or other links.
  • the original voice data is the voice data collected in real time when the parties to the trial respond to the trial prompt text.
  • the judge will guide or ask questions to the parties (such as the plaintiff, lawyer or witness) in the court trial based on the trial prompt information.
  • the parties in the court trial need to respond.
  • the voice data collected by the microphone is Raw voice data.
  • the intelligent auxiliary trial system includes at least one microphone connected to the server, and each microphone corresponds to a microphone identifier, and the microphone identifier is an identifier for uniquely identifying different microphones. When the microphone collects raw voice data in real time, its raw voice data is associated with the corresponding microphone identifier.
  • S402 Perform voiceprint recognition on the original voice data, obtain a voiceprint recognition result, and determine an object identifier corresponding to the original voice data according to the voiceprint recognition result.
  • the voiceprint recognition is performed on the original voice data to obtain the voiceprint recognition result.
  • the server uses the preset voiceprint feature extraction algorithm to extract the voiceprint feature of the original voice data, and then performs the voiceprint feature extraction based on the extracted voiceprint feature. Pattern recognition to determine the identity of the speaker corresponding to the original voice data.
  • the object identifier is used to uniquely identify different parties in the court hearing.
  • the server uses a voiceprint feature extraction algorithm to extract voiceprint features from the original voice data, obtains the target voiceprint feature, and determines whether there is a standard voiceprint feature corresponding to the target voiceprint feature to obtain the corresponding voiceprint recognition result .
  • the voiceprint recognition result includes two results of the existence of a standard voiceprint feature and the absence of a standard voiceprint feature.
  • step S402 specifically includes the following steps:
  • S4011 Use the voiceprint feature extraction algorithm to extract the voiceprint feature of the original voice data, obtain the target voiceprint feature, and determine whether there is a standard voiceprint feature corresponding to the target voiceprint feature.
  • the voiceprint feature extraction algorithm is an algorithm used to extract voiceprint features from voice data to determine the voiceprint features corresponding to the original voice data.
  • the voiceprint feature extraction algorithm includes but is not limited to the MFCC extraction algorithm, and the extracted target voiceprint feature is the MFCC feature.
  • MFCC Mel-scale Frequency Cepstral Coefficients
  • the Mel scale describes the non-linear characteristics of the human ear frequency.
  • the target voiceprint feature is the voiceprint feature extracted from the original voice data.
  • the standard voiceprint feature is the voiceprint feature extracted based on the voice data of the parties to the trial collected during the pre-trial preparation process.
  • the standard voiceprint feature is also the MFCC feature extracted by the MFCC extraction algorithm.
  • the standard voiceprint features collected by the parties in the court trial and their corresponding identification are stored in a database in association with each other for subsequent identification processing.
  • the server uses the cosine similarity algorithm or other similarity algorithms to calculate the similarity between the target voiceprint feature and each standard voiceprint feature pre-stored in the server to obtain the target similarity; if the target similarity is greater than the preset If the similarity threshold is the same, it is determined that there is a standard voiceprint feature; if the target similarity is not greater than the preset similarity threshold, it is determined that there is no standard voiceprint feature.
  • the preset similarity threshold is a threshold used to evaluate whether the similarity reaches the criteria for identifying the same speaker.
  • the object identifier corresponding to the original voice data is determined according to the identity tag corresponding to the standard voiceprint feature, so as to quickly determine the original voiceprint feature.
  • Object ID corresponding to the voice data The identity mark is used to distinguish the identity of the parties in the court hearing, such as the plaintiff, the court, and the witness.
  • the object identifier is used to uniquely identify different parties in the court hearing.
  • the object identification can be a serial number identification added to the identity identification. For example, in the case of multiple plaintiffs, the object identification in the form of plaintiff 01 and plaintiff 02 can be used to distinguish.
  • the serial number identification can be determined according to the sequence of collecting standard identity features during the pre-trial preparation process, or according to the sequence of their speeches during the court hearing, so that each party to the court hearing has an object identifier that uniquely identifies its identity.
  • the server does not store the standard voiceprint feature corresponding to the target voiceprint feature, it means that the speaker has not collected the standard voiceprint feature in advance during the pre-court preparation process. At this time, it can be based on the original voice data.
  • the carried microphone ID queries the microphone information table, obtains the ID ID corresponding to the microphone ID, and generates the corresponding object ID based on the ID ID.
  • the microphone information table is an information comparison table used to determine the identity identifier corresponding to the speaker according to the placement position of the microphone, and the microphone information table associates the microphone identifier with its corresponding identity identifier.
  • the identity corresponding to the microphone identity is the witness, and the corresponding object identity is generated based on the identity. Specifically, it refers to the serial number identity formed by adding the identity of the witness and the order of speaking. Obtain the corresponding object ID, such as witness 01, witness 02, etc.
  • the target voiceprint feature extracted from the original voice data is used to determine whether there is a voiceprint recognition result corresponding to the standard voiceprint feature to determine whether it is based on the standard voiceprint feature or the microphone
  • the identifier determines the corresponding object identifier to ensure the uniqueness of the determined identity identifier.
  • S403 Perform text translation on the original voice data, obtain the original text data corresponding to the object identifier, and store the object identifier and the original text data in a corresponding position in the standard court trial transcript template.
  • the text translation of the original voice data refers to the process of translating the original voice data into data in text form.
  • the original text data refers to the text data translated from the original voice data.
  • the server can use, but is not limited to, a static decoding network to perform text translation on the original voice data. Since the static decoding network has fully expanded the search space, the decoding speed is very fast when performing text translation, which can be fast Obtain the original text data corresponding to the object ID. Understandably, the server receives the original voice data collected by the microphone, and then uses the static decoding network to translate the original voice data to quickly obtain the corresponding original text data without manual input by the clerk, thereby speeding up the original text data Input efficiency.
  • the server performs text translation on the original voice data to obtain the corresponding original text data
  • the original text data and its corresponding object identifier are associated and stored in the corresponding position of the standard court trial transcript template, that is, the original text data is filled in The position corresponding to the object identifier in the current trial link of the standard court trial transcript template.
  • the original voice data is the voice data that responds to the trial prompt text "Does the plaintiff have any supplements to your prosecution?" and the corresponding object is identified as the plaintiff, the original voice data can be translated into the original
  • the text data is filled in the position corresponding to the trial prompt text in the standard court trial transcript template to improve the input efficiency of the original text data and reduce the work burden of the clerks.
  • S404 Query the prior text database based on the original text data, and determine whether there is prior text data corresponding to the original text data.
  • the previous text data refers to the text data that has been formed and recorded in the corresponding position of the standard court trial transcript template before the original voice data is collected by the microphone. Since the court trial is a process in which the parties to the trial play a game against the same event, during the trial, the parties to the trial will discuss the same event from different perspectives. The content of the discussion may be relevant.
  • the previous text data can be understood as The text content corresponding to the content discussed in the original voice data before the original voice data is collected. Take time as an example. In a criminal case, the time node of the case is a key factor that affects the evidence chain of the case formation or the severity of the sentence. The plaintiff, lawyer, and witnesses may publish different original voice data based on these time nodes.
  • various time nodes related to the pending case are the key factors affecting the determination of liability for breach of contract.
  • the plaintiff, the lawyer and the witnesseses may publish different original voice data based on these time nodes, and determine the previous text data and the original text data according to the order of their formation time.
  • step S404 specifically includes the following steps:
  • S4041 Use a keyword extraction algorithm to extract keywords from the original text data to obtain text keywords.
  • text keywords are keywords extracted from the original text data.
  • Keyword extraction algorithm is an algorithm used to extract keywords from text data.
  • keyword extraction algorithms such as TextRank, LDA, TPR-TextRank, etc. are used, but not limited to, to perform keyword extraction on the original text data to obtain text keywords corresponding to the original text data.
  • S4042 Query the thesaurus based on the text keywords, and obtain text synonyms corresponding to the text keywords.
  • the thesaurus is used to store a pre-set database for storing synonym relationships.
  • Text synonyms are synonyms recorded in the thesaurus that have a synonym relationship with text keywords.
  • synonym groups with synonym relationships are pre-stored in the thesaurus. These synonym groups may specifically be synonym groups involved in the trial process of the case, so that the server can query the text keywords extracted from the original text data. Corresponding text keywords to help expand the scope of the query in the subsequent query process.
  • S4043 Query the prior text database according to the text keywords and text synonyms, and determine whether there is prior text data containing the text keywords or text synonyms.
  • the previous text database is a database used to store all previous text data formed before the original voice data is collected.
  • the prior text database is queried according to text keywords and text synonyms, and it is determined whether there is prior text data corresponding to the text keyword in the prior text database, or whether there is a prior text data corresponding to the text synonym. First text data to expand the search range of previous text data.
  • the speech data is translated into prior text data and stored in the prior text database for subsequent semantic analysis based on the original text data and prior text data To determine whether the two expressions have the same meaning, that is, the step S405 is subsequently executed.
  • S405 If prior text data exists, perform semantic analysis on the original text data and prior text data to determine the semantic analysis result, highlight the original text data based on the semantic analysis result, and display the trial corresponding to the semantic analysis result Prompt text, repeated execution to obtain the original voice data collected by the microphone and corresponding to the trial prompt text.
  • semantic analysis is performed on the original text data and the prior text data to determine the semantic analysis result, which specifically includes: using a semantic analysis tool to analyze the original document data and the previous text data. Perform semantic analysis on the first text data to determine whether the semantics of the original document data and the previous text data are the same or different, and obtain the corresponding semantic analysis results.
  • the speech analysis results include the same semantic analysis results and different semantic analysis results.
  • the semantic analysis tool may adopt, but is not limited to, an analysis tool created by NLP (Natural Language Processing) technology.
  • the prior text data corresponding to party A in the court trial records "I bought a batch of products worth 100,000 from B on March 10", while the original text data corresponding to party B in the court trial records "I was on March 10 Sell a batch of products with a value of 100,000 to A”.
  • the parties involved, time, subject matter, and price are all the same.
  • semantic analysis tools to analyze the original document data and the previous text
  • the previous text data corresponding to party A in the court trial recorded "I purchased a batch of products worth 100,000 from B on March 10”
  • the original text data corresponding to party B in the court trial recorded "I was on March 8. No. Selling a batch of products with a value of 150,000 to A”.
  • the parties involved and the subject matter are the same, the time and price are not the same, and the opinions described by the two are determined Different, obtain different semantic analysis results.
  • the original text data is highlighted according to the semantic analysis result, and the trial prompt text corresponding to the semantic analysis result is displayed, which specifically refers to whether the original text data and the previous text data have the same semantics or different semantics according to the semantic analysis result , So as to determine whether the original text data corresponding to the current trial link is highlighted in different ways in response to the facts determined by the two parties, the focus of the dispute and whether the description is inconsistent, so that the judge can highlight the processing results during the trial Understanding the above situation will help reduce the workload of the judge in the trial process, thereby reducing the workload, and display the trial prompt information corresponding to the semantic analysis result, which will help speed up the trial progress of the court trial.
  • step S405 specifically includes the following steps:
  • the semantic analysis result is the same, and the object identifier corresponding to the original text data and the previous text data is the same identifier, it means that the speaker who spoke the original text data and the previous text data is the same person, and the original text data and The semantics of the previous text data is the same, and there is no conflict between the previous and the next. It can be understood as a discussion without special circumstances. Therefore, the original text data is not highlighted, and the next trial link corresponding to the standard court trial transcript template is displayed. For the corresponding trial prompt text, repeat the steps of acquiring the original voice data corresponding to the trial prompt text collected by the microphone in step S401 and the subsequent steps.
  • the first highlighting mode is a preset mode for highlighting the inconsistent content discussed by the same speaker.
  • the first highlighting mode can adopt font color, background color, bold, slanted or underlined, etc. The form is highlighted.
  • the semantic analysis results are different, and the object identifiers corresponding to the original text data and the previous text data are the same, indicating that the speaker who spoke the original text data and the previous text data is the same person, and the original text data and The semantics of the previous text data are different, and there is a problem of contradictory statements. It is very likely that the product was produced because of the special situation of the speaker lying. Therefore, the first highlighting mode needs to be used to highlight the original text data.
  • the second highlighting mode is used to highlight the original text data, and display the information including the non-dispute prompt information Trial reminder text.
  • the second highlighting mode is a preset mode for highlighting content that is not contradictory to different speakers’ discussions. Understandably, the second highlighting mode is a mode different from the first highlighting mode. It can be highlighted in the form of font color, background color, bold, slanted, or underlined.
  • the second highlighting mode is required to highlight the original text data. Displaying the trial reminder text that includes non-controversial reminder information can help determine the undisputed facts in the court trial process, and better control the guiding issues in the court trial process, helping to reduce the workload of the judge in the trial process, thereby Reduce the workload.
  • the third highlighting mode is a preset mode for highlighting conflicting content discussed by different speakers. Understandably, this third highlighting mode is the same as the previous first highlighting mode and second highlighting mode. Modes with different display modes can also be highlighted in the form of font color, background color, bold, oblique, or underlined.
  • the third highlight mode can be used to highlight the original text data.
  • Display the trial prompt text containing the prompt information of the dispute focus which helps to determine the focus of the dispute in the court hearing process, and then better control the guiding problems in the court hearing process, which helps to reduce the workload of the judge during the trial process, thereby reducing Work load.
  • the server can also use the voice polygraph model preset on the server to process the original voice data to obtain the probability of lying. If the probability of lying is greater than With the preset probability threshold, the lie display mode is used to highlight the original text data, so that the judge knows whether the parties in the trial are lying in a timely manner during the trial, so as to ensure a fair and just trial of the case.
  • the voice test model may be a model used on a voice tester currently publicly available on the market, so as to determine the probability that the original voice data spoken by the speaker is a lie based on the voice frequency or voice tone contained in the original voice data .
  • the preset probability threshold is a preset threshold used to evaluate whether the probability of determining a lie is reached.
  • the lie display mode is a preset mode for highlighting the original text data with a higher probability of lying.
  • the standard court trial record template can be used to determine whether there is a trial prompt text corresponding to the next trial link; if there is a trial corresponding to the next trial link
  • prompting text repeat the steps to obtain the original voice data corresponding to the trial prompt text collected by the microphone and the subsequent steps (ie steps S402, S403); if there is no trial prompt text corresponding to the next trial session, the court is deemed to be heard
  • a court trial record file is formed based on all the original text data filled in the corresponding position in the standard court trial record template, and the court trial record file is stored in the database so that the judge can make a judgment document based on the court trial record file.
  • the intelligent assisted trial method provided in this embodiment, after the original voice data corresponding to the trial prompt text corresponding to the current trial session is collected through a microphone in real time, it is determined according to the voiceprint recognition result of the voiceprint recognition of the original voice data Object identification to determine the identity of the speaker corresponding to the original voice data; the original text data obtained by the text translation of the original voice data and the object identification are stored in the corresponding position of the standard court trial transcript template, thereby improving the input efficiency of the original text data , No need for clerks to enter verbatim, reducing the work burden of clerks.
  • the original text data is highlighted, and the trial prompt text corresponding to the semantic analysis result is displayed , which enables judges to understand the special situations corresponding to different semantic analysis results according to the highlighted processing results during the court trial, which helps to reduce the workload of the judges during the court trial, thereby reducing the workload and displaying the trial corresponding to the semantic analysis result
  • the prompt information helps to speed up the trial progress of the court trial and improve the efficiency of the trial.
  • an intelligent auxiliary trial device is provided, and the intelligent auxiliary trial device corresponds one-to-one with the intelligent auxiliary trial method in the foregoing embodiment.
  • the detailed description of each functional module of the intelligent auxiliary trial device is as follows:
  • the case description information obtaining module 501 is used to obtain the court trial transcript file corresponding to the case to be judged from the database, and extract the case description information from the court trial transcript file;
  • the target word segmentation result acquisition module 502 is configured to use word segmentation tools to segment the case description information to obtain a target word segmentation result, and the target word segmentation result includes multiple target word segmentation;
  • the target keyword determining module 503 is configured to query a keyword database based on each target word segmentation, and determine the original keywords stored in the keyword database that match the target word segmentation as the target keyword;
  • the target case type determination module 504 is configured to query the case type information database according to the target keyword, and obtain at least one target case type that matches the target keyword;
  • the target push law acquisition module 505 acquires prior knowledge matching at least one target case type from the prior knowledge base, and obtains the target push law based on the prior knowledge;
  • the recommended trial viewpoint acquisition module 506 is used to perform semantic analysis on case description information using a semantic analysis model to obtain standardized information, query the trial viewpoint database based on the standardized information, and obtain corresponding recommended trial viewpoints;
  • the trial suggestion file acquisition module 507 is used to push the law and recommended trial opinions according to the target, and acquire and display the trial suggestion file.
  • the intelligent auxiliary trial device further includes:
  • the original voice data acquisition module is used to display the trial prompt text corresponding to the current trial link in the standard court trial transcript template on the client, and obtain the original voice data collected by the microphone and corresponding to the trial prompt text;
  • the object identification acquisition module is used to perform voiceprint recognition on the original voice data, obtain the voiceprint recognition result, and determine the object identifier corresponding to the original voice data according to the voiceprint recognition result;
  • the original text data acquisition module is used to translate the original speech data, obtain the original text data corresponding to the object identification, and store the object identification and the original text data in the corresponding position of the standard court trial transcript template;
  • the prior text data judgment module is used to query the prior text database based on the original text data to determine whether there is prior text data corresponding to the original text data;
  • the highlight processing module is used to perform semantic analysis on the original text data and previous text data if there is prior text data, determine the semantic analysis result, and perform highlight processing on the original text data according to the semantic analysis result, display and semantic analysis
  • the trial prompt text corresponding to the result is repeatedly executed to obtain the original voice data corresponding to the trial prompt text collected by the microphone;
  • the transcript file acquisition module is used to repeatedly execute the trial prompt text corresponding to the next trial link in the standard court trial transcript template if there is no prior text data, and obtain the corresponding trial prompt text collected by the microphone The original voice data, until there is no trial prompt text corresponding to the next trial link, obtain the court trial transcript file and store the court trial transcript file in the database.
  • the object identification acquisition module includes:
  • the voiceprint feature extraction and judgment unit is used to perform voiceprint feature extraction on the original voice data by using the voiceprint feature extraction algorithm, obtain the target voiceprint feature, and determine whether there is a standard voiceprint feature corresponding to the target voiceprint feature;
  • the first object identification determining unit is configured to determine the object identification corresponding to the original voice data according to the identity identification corresponding to the standard voiceprint feature if there is a standard voiceprint feature;
  • the second object identifier determining unit is configured to determine the object identifier corresponding to the original voice data according to the microphone identifier corresponding to the original voice data if there is no standard voiceprint feature.
  • the prior text data judgment module includes:
  • the text keyword acquisition unit is used to extract keywords from the original text data using a keyword extraction algorithm to obtain text keywords;
  • the text synonym acquisition unit is used to query the thesaurus based on text keywords to obtain text synonyms corresponding to the text keywords;
  • the prior text query judgment unit is used to query the prior text database based on text keywords and text synonyms, and determine whether there is prior text data containing text keywords or text synonyms;
  • the first judgment processing unit is configured to determine that there is prior text data corresponding to the original text data if there is prior text data containing text keywords or text synonyms;
  • the second judgment processing unit is configured to determine that there is no prior text data corresponding to the original text data if there is no prior text data containing text keywords or text synonyms.
  • the highlight processing module includes:
  • the first display processing unit is used for if the semantic analysis result is the same, and the original text data and the object identifier corresponding to the previous text data are the same identifier, the original text data is not highlighted, and the standard court trial record template is displayed. Trial prompt text corresponding to the next trial session;
  • the second display processing unit is used to perform highlight processing on the original text data using the first highlight mode if the semantic analysis results are different, and the original text data and the object identifier corresponding to the previous text data are the same identifier, and display Trial reminder text including contradictory reminder information;
  • the third display processing unit is configured to use the second highlighting mode to highlight the original text data if the semantic analysis results are the same, and the original text data and the object identifier corresponding to the previous text data are not the same identifier, and display Trial reminder text including no dispute reminder information;
  • the fourth display processing unit is used to perform highlight processing on the original text data using the third highlighting mode if the semantic analysis results are different and the object identifiers corresponding to the original text data and the previous text data are not the same identifier, Display the trial reminder text including the reminder of the focus of the dispute.
  • the target word segmentation result acquisition module includes:
  • the text segmentation processing unit is used to use the search engine mode of the stuttering word segmentation tool to segment the case description information to obtain the text segmentation result.
  • the text segmentation result includes N first-level word segmentation;
  • the first optimization processing unit is configured to: if the superposition of consecutive k-1 first-level participles in any continuous k first-level participles is equal to the k-th first-level participle, and there are at least two consecutive k-1 first-level participles The combination of graded participles is equal to the k-th first-level participle, and only at least two first-level participles that are combined and equal to the k-th first-level participle are retained as target participles to obtain the target segmentation result;
  • the second optimization processing unit is configured to: if the superposition of k-1 consecutive first-level participles in any continuous k first-level participles is equal to the k-th first-level participle, and there are no at least two one in any continuous k first-level participles The combination of level participle is equal to the k-th first-level participle, then delete the first k-1 first-level participles, keep the k-th first-level participle as the target participle, and obtain the target participle result;
  • the third optimization processing unit is used to delete the k-th first-level participle and keep the first k-1 if the combination of k-1 first-level participles in any continuous k first-level participles is equal to the k-th first-level participle
  • the first-level word segmentation is used as the target word segmentation to obtain the target word segmentation result.
  • the target keyword determination module includes:
  • the target synonym obtaining unit is used to query the thesaurus according to the target word segmentation, and obtain the target synonym corresponding to the target word segmentation;
  • the target keyword determining unit is used to query the keyword database according to the target word segmentation and target synonyms, and determine whether there is an original keyword that matches the target word segmentation or target synonym. If the original keyword exists, the original keyword is determined as the target key word.
  • the intelligent auxiliary trial device further includes:
  • the update task acquisition module is used to acquire the data update task, which includes the original case type and task update time;
  • the first data acquisition module is used to determine the change time if the judgment basis corresponding to the original case type is changed after the task update time, and obtain the history corresponding to the original case type between the change time and the current system time Judgment data, determine the data to be processed based on historical judgment data;
  • the second data acquisition module is used to obtain historical judgment data corresponding to the original case type in the preset period before the current time of the system if the judgment basis corresponding to the original case type does not change after the task update time , Determine the data to be processed based on historical judgment data;
  • a priori knowledge base building module used to extract prior information from the data to be processed, and build a priori knowledge base corresponding to the original case type based on the prior information
  • the trial viewpoint database building module is used to extract historical description information and historical trial viewpoints from the data to be processed, and construct a trial viewpoint database corresponding to the original case type based on the historical description information and historical trial viewpoints.
  • the data to be processed includes the target area;
  • the historical judgment data corresponding to the same target area and the original case type is determined as the data to be processed
  • the historical judgment data corresponding to the same original case type is determined as the data to be processed.
  • Each module in the above-mentioned intelligent auxiliary trial device can be implemented in whole or in part by software, hardware and a combination thereof.
  • the foregoing modules may be embedded in the form of hardware or independent of the processor in the computer device, or may be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the foregoing modules.
  • a computer device is provided.
  • the computer device may be a server, and its internal structure diagram may be as shown in FIG. 6.
  • the computer equipment includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide calculation and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium stores an operating system, computer readable instructions, and a database.
  • the internal memory provides an environment for the operation of the operating system and computer-readable instructions in the non-volatile storage medium.
  • the database of the computer equipment is used to store data used or generated during the execution of the intelligent auxiliary trial method.
  • the network interface of the computer device is used to communicate with an external terminal through a network connection.
  • the computer-readable instruction is executed by the processor to realize an intelligent auxiliary trial method.
  • a computer device including a memory, a processor, and computer-readable instructions stored in the memory and capable of running on the processor.
  • the processor executes the computer-readable instructions to realize the intelligence in the above-mentioned embodiments.
  • the auxiliary trial method such as shown in Figure 2- Figure 4, in order to avoid repetition, will not be repeated here.
  • the processor executes the computer-readable instructions, the functions of the modules/units of the above-mentioned intelligent auxiliary trial device are realized.
  • the modules are shown in FIG.
  • one or more non-volatile readable storage media storing computer readable instructions are provided.
  • the computer readable storage medium stores computer readable instructions, and the computer readable instructions are stored by one or more
  • the one or more processors are executed to implement the intelligent auxiliary judgment method in the above-mentioned embodiment, for example, as shown in Figs. 2 to 4, in order to avoid repetition, details are not repeated here.
  • the functions of the modules/units of the above-mentioned intelligent auxiliary trial device when the computer-readable instructions are executed by the processor are the modules shown in FIG. 5, and to avoid repetition, the description will be different here.
  • Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • ROM read only memory
  • PROM programmable ROM
  • EPROM electrically programmable ROM
  • EEPROM electrically erasable programmable ROM
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • General Health & Medical Sciences (AREA)
  • Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Technology Law (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Machine Translation (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

L'invention concerne un procédé et un appareil de jugement auxiliaire intelligent, ainsi qu'un dispositif informatique et un support de stockage. Le procédé consiste à : acquérir un fichier de transcription de procès correspondant à une affaire à juger, et extraire des informations de description d'affaire du fichier de transcription de procès (S201) ; effectuer une segmentation en mots sur les informations de description d'affaire au moyen d'un outil de segmentation en mots pour acquérir un résultat de segmentation en mots cibles (S202) ; interroger une base de mots-clés sur la base de chaque mot segmenté cible, et déterminer un mot-clé cible (S203) ; interroger une base d'informations de types d'affaires en fonction du mot-clé cible, et acquérir au moins un type d'affaire cible concordant (S204) ; acquérir, à partir d'une base de connaissances a priori, une connaissance a priori concordant avec l'au moins un type d'affaire cible, et traiter la connaissance a priori au moyen d'un modèle de pousser intelligent de disposition légale pour acquérir une disposition légale poussée cible (S205) ; effectuer une analyse sémantique sur les informations de description d'affaire au moyen d'un modèle d'analyse sémantique pour acquérir des points de vue de jugement recommandés correspondants (S206) ; et acquérir et afficher, en fonction de la disposition légale poussée cible et des points de vue de jugement recommandés, un fichier de suggestion de jugement (S207), de façon à réduire la charge de travail de juges pendant un processus de jugement, ce qui permet de réduire la pénibilité du travail.
PCT/CN2019/116487 2019-08-16 2019-11-08 Procédé et appareil de jugement auxiliaire intelligent, et dispositif informatique et support de stockage WO2021031383A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910756628.7 2019-08-16
CN201910756628.7A CN110675288B (zh) 2019-08-16 2019-08-16 智能辅助审判方法、装置、计算机设备及存储介质

Publications (1)

Publication Number Publication Date
WO2021031383A1 true WO2021031383A1 (fr) 2021-02-25

Family

ID=69075331

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/116487 WO2021031383A1 (fr) 2019-08-16 2019-11-08 Procédé et appareil de jugement auxiliaire intelligent, et dispositif informatique et support de stockage

Country Status (2)

Country Link
CN (1) CN110675288B (fr)
WO (1) WO2021031383A1 (fr)

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112884440A (zh) * 2021-03-02 2021-06-01 岭东核电有限公司 核电试验中的试验工序执行方法、装置和计算机设备
CN113570269A (zh) * 2021-08-03 2021-10-29 工银科技有限公司 运维项目的管理方法、装置、设备、介质和程序产品
CN113901125A (zh) * 2021-09-17 2022-01-07 盐城金堤科技有限公司 数据展示方法和装置、及存储介质和电子设备
CN114185665A (zh) * 2021-12-16 2022-03-15 北京少广科技集团有限公司 一种云计算操作系统高效任务调度技术方法
CN114416988A (zh) * 2022-01-17 2022-04-29 国网福建省电力有限公司 基于自然语言处理的缺陷自动评级及处置建议推送方法
CN116596709A (zh) * 2023-07-19 2023-08-15 北京分音塔科技有限公司 一种辅助审判方法、装置、设备及存储介质
CN116758947A (zh) * 2023-08-14 2023-09-15 北京分音塔科技有限公司 基于音频情感的辅助审判方法、装置、设备及存储介质
CN117009605A (zh) * 2023-08-08 2023-11-07 四川大学 一种策略化创新设计问题求解方法及系统
CN117035406A (zh) * 2023-07-31 2023-11-10 北京华夏电通科技股份有限公司 审判流程智能管控方法、装置和设备
CN117371916A (zh) * 2023-12-05 2024-01-09 智粤铁路设备有限公司 基于数字化维护的数据处理方法及工量具智能管理系统
CN117473074A (zh) * 2023-11-01 2024-01-30 中国通信建设集团有限公司数智科创分公司 一种基于人工智能的司法案例智能信息匹配系统及方法
CN117609487A (zh) * 2024-01-19 2024-02-27 武汉百智诚远科技有限公司 基于人工智能的法律条文快速检索方法及系统
CN118673893A (zh) * 2024-08-26 2024-09-20 浙江海规技术有限公司 一种文书生成方法、装置、设备及计算机可读存储介质

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111563107A (zh) * 2020-05-25 2020-08-21 泰康保险集团股份有限公司 信息推荐的方法、装置、电子设备和存储介质
CN111930933A (zh) * 2020-05-29 2020-11-13 深圳壹账通智能科技有限公司 一种基于人工智能的检务案件处理方法及装置
CN111709236B (zh) * 2020-05-29 2024-01-09 中山大学 一种基于案件相似度匹配的审判风险预警方法
CN111859888B (zh) * 2020-07-22 2024-04-02 北京致医健康信息技术有限公司 一种诊断辅助方法、装置、电子设备和存储介质
CN112559748A (zh) * 2020-12-18 2021-03-26 厦门市法度信息科技有限公司 一种笔录数据案由分类方法、终端设备及存储介质
CN112765974B (zh) * 2021-01-19 2023-11-24 卡奥斯数字科技(北京)有限公司 一种业务辅助方法、电子设备及可读存储介质
CN113157722B (zh) * 2021-04-01 2023-12-26 北京达佳互联信息技术有限公司 一种数据处理方法、装置、服务器、系统及存储介质
CN113283765A (zh) * 2021-05-31 2021-08-20 浙江环玛信息科技有限公司 智慧法院案件数据处理方法及系统
CN113609256B (zh) * 2021-08-05 2022-03-15 郑州银丰电子科技有限公司 基于大数据的智慧法院管理系统和方法
CN114492436B (zh) * 2022-02-11 2024-07-26 国家电网有限公司华东分部 一种审计访谈信息的处理方法、装置和系统
CN116244315B (zh) * 2022-12-08 2023-11-10 南京擎盾信息科技有限公司 一种法律法规数据库时效性动态更新的方法和系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090070099A1 (en) * 2006-10-10 2009-03-12 Konstantin Anisimovich Method for translating documents from one language into another using a database of translations, a terminology dictionary, a translation dictionary, and a machine translation system
CN107633465A (zh) * 2017-08-21 2018-01-26 厦门能见易判信息科技有限公司 智能辅助判案方法
CN107818138A (zh) * 2017-09-28 2018-03-20 银江股份有限公司 一种案件法律条例推荐方法及系统
CN108009299A (zh) * 2017-12-28 2018-05-08 北京市律典通科技有限公司 法律审判业务处理方法和装置
CN108595532A (zh) * 2018-04-02 2018-09-28 三峡大学 一种法律文本的量子聚类系统及方法
CN109002431A (zh) * 2018-07-20 2018-12-14 吴怡 法律意见书生成系统及生产方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
AU2009201864A1 (en) * 2009-05-11 2010-11-25 Fiona MacPhee Semi-automated court document production
CN109192213B (zh) * 2018-08-21 2023-10-20 平安科技（深圳）有限公司 庭审语音实时转写方法、装置、计算机设备及存储介质

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090070099A1 (en) * 2006-10-10 2009-03-12 Konstantin Anisimovich Method for translating documents from one language into another using a database of translations, a terminology dictionary, a translation dictionary, and a machine translation system
CN107633465A (zh) * 2017-08-21 2018-01-26 厦门能见易判信息科技有限公司 智能辅助判案方法
CN107818138A (zh) * 2017-09-28 2018-03-20 银江股份有限公司 一种案件法律条例推荐方法及系统
CN108009299A (zh) * 2017-12-28 2018-05-08 北京市律典通科技有限公司 法律审判业务处理方法和装置
CN108595532A (zh) * 2018-04-02 2018-09-28 三峡大学 一种法律文本的量子聚类系统及方法
CN109002431A (zh) * 2018-07-20 2018-12-14 吴怡 法律意见书生成系统及生产方法

Cited By (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112884440A (zh) * 2021-03-02 2021-06-01 岭东核电有限公司 核电试验中的试验工序执行方法、装置和计算机设备
CN112884440B (zh) * 2021-03-02 2024-05-24 岭东核电有限公司 核电试验中的试验工序执行方法、装置和计算机设备
CN113570269A (zh) * 2021-08-03 2021-10-29 工银科技有限公司 运维项目的管理方法、装置、设备、介质和程序产品
CN113901125A (zh) * 2021-09-17 2022-01-07 盐城金堤科技有限公司 数据展示方法和装置、及存储介质和电子设备
CN114185665A (zh) * 2021-12-16 2022-03-15 北京少广科技集团有限公司 一种云计算操作系统高效任务调度技术方法
CN114416988A (zh) * 2022-01-17 2022-04-29 国网福建省电力有限公司 基于自然语言处理的缺陷自动评级及处置建议推送方法
CN116596709B (zh) * 2023-07-19 2024-02-06 北京分音塔科技有限公司 一种辅助审判方法、装置、设备及存储介质
CN116596709A (zh) * 2023-07-19 2023-08-15 北京分音塔科技有限公司 一种辅助审判方法、装置、设备及存储介质
CN117035406A (zh) * 2023-07-31 2023-11-10 北京华夏电通科技股份有限公司 审判流程智能管控方法、装置和设备
CN117009605A (zh) * 2023-08-08 2023-11-07 四川大学 一种策略化创新设计问题求解方法及系统
CN117009605B (zh) * 2023-08-08 2024-04-02 四川大学 一种策略化创新设计问题求解方法及系统
CN116758947B (zh) * 2023-08-14 2023-10-20 北京分音塔科技有限公司 基于音频情感的辅助审判方法、装置、设备及存储介质
CN116758947A (zh) * 2023-08-14 2023-09-15 北京分音塔科技有限公司 基于音频情感的辅助审判方法、装置、设备及存储介质
CN117473074A (zh) * 2023-11-01 2024-01-30 中国通信建设集团有限公司数智科创分公司 一种基于人工智能的司法案例智能信息匹配系统及方法
CN117371916A (zh) * 2023-12-05 2024-01-09 智粤铁路设备有限公司 基于数字化维护的数据处理方法及工量具智能管理系统
CN117371916B (zh) * 2023-12-05 2024-02-23 智粤铁路设备有限公司 基于数字化维护的数据处理方法及工量具智能管理系统
CN117609487A (zh) * 2024-01-19 2024-02-27 武汉百智诚远科技有限公司 基于人工智能的法律条文快速检索方法及系统
CN117609487B (zh) * 2024-01-19 2024-04-09 武汉百智诚远科技有限公司 基于人工智能的法律条文快速检索方法及系统
CN118673893A (zh) * 2024-08-26 2024-09-20 浙江海规技术有限公司 一种文书生成方法、装置、设备及计算机可读存储介质

Also Published As

Publication number Publication date
CN110675288A (zh) 2020-01-10
CN110675288B (zh) 2023-07-14

Similar Documents

Publication Publication Date Title
WO2021031383A1 (fr) Procédé et appareil de jugement auxiliaire intelligent, et dispositif informatique et support de stockage
CN110704571B (zh) 庭审辅助处理方法、审判辅助处理方法、装置、设备及介质
CN110874531B (zh) 一种话题分析方法、装置和存储介质
Hassan et al. The quest to automate fact-checking
CN115238101A (zh) 一种面向多类型知识库的多引擎智能问答系统
WO2021000497A1 (fr) Procédé et appareil de récupération, dispositif informatique et support de stockage
KR102216768B1 (ko) 심리상담 데이터를 이용한 텍스트 내 감정분석을 위한 시스템 및 방법
TWI650719B (zh) 從文字內容評核客戶服務品質之系統及方法
WO2020181808A1 (fr) Procédé et appareil de prévision de ponctuation de texte, et dispositif informatique et support de stockage
CN113094578B (zh) 基于深度学习的内容推荐方法、装置、设备及存储介质
WO2020010834A1 (fr) Procédé, appareil et dispositif de généralisation de bibliothèque de questions et de réponses d'une foire aux questions
WO2019232893A1 (fr) Procédé et dispositif d'analyse d'une émotion dans un texte, appareil informatique et support de stockage
CN106601237A (zh) 交互式语音应答系统及其语音识别方法
US12093648B2 (en) Systems and methods for producing a semantic representation of a document
CN105912629A (zh) 一种智能问答方法及装置
CN105787134B (zh) 智能问答方法、装置及系统
CN109710918B (zh) 舆情关系识别方法、装置、计算机设备和存储介质
CN110309504B (zh) 基于分词的文本处理方法、装置、设备及存储介质
CN117743590B (zh) 一种基于大语言模型的法律辅助方法及系统
CN110347812A (zh) 一种面向司法文本的搜索排序方法及系统
CN112347121A (zh) 一种可配置的自然语言转sql的方法及系统
CN111145053A (zh) 基于人工智能的企业法律顾问管理系统及方法
CN112668284B (zh) 一种法律文书分段方法及系统
CN113590768B (zh) 一种文本关联度模型的训练方法及装置、问答方法及装置
CN105786929A (zh) 一种信息监测方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19942209

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19942209

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 16/08/2022)

122 Ep: pct application non-entry in european phase

Ref document number: 19942209

Country of ref document: EP

Kind code of ref document: A1