CN111797230A - Legal three-layer theory automatic reasoning method and device and electronic equipment - Google Patents
Legal three-layer theory automatic reasoning method and device and electronic equipment Download PDFInfo
- Publication number
- CN111797230A CN111797230A CN202010532799.4A CN202010532799A CN111797230A CN 111797230 A CN111797230 A CN 111797230A CN 202010532799 A CN202010532799 A CN 202010532799A CN 111797230 A CN111797230 A CN 111797230A
- Authority
- CN
- China
- Prior art keywords
- legal
- event
- reasoning
- proposition
- requirement
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 59
- 230000003993 interaction Effects 0.000 claims abstract description 87
- 238000012549 training Methods 0.000 claims abstract description 10
- 230000015654 memory Effects 0.000 claims description 17
- 230000002452 interceptive effect Effects 0.000 claims description 15
- 238000010276 construction Methods 0.000 claims description 13
- 238000009826 distribution Methods 0.000 claims description 9
- 239000013598 vector Substances 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 3
- 238000013139 quantization Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 description 14
- 230000009471 action Effects 0.000 description 10
- 238000010586 diagram Methods 0.000 description 10
- 238000003860 storage Methods 0.000 description 9
- 238000012545 processing Methods 0.000 description 5
- LFQSCWFLJHTTHZ-UHFFFAOYSA-N Ethanol Chemical compound CCO LFQSCWFLJHTTHZ-UHFFFAOYSA-N 0.000 description 4
- 230000006399 behavior Effects 0.000 description 4
- 238000001228 spectrum Methods 0.000 description 4
- 230000001502 supplementing effect Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 239000000470 constituent Substances 0.000 description 3
- 230000006378 damage Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 239000007787 solid Substances 0.000 description 3
- 230000004888 barrier function Effects 0.000 description 2
- 239000003795 chemical substances by application Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000003058 natural language processing Methods 0.000 description 2
- 230000000750 progressive effect Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 208000034656 Contusions Diseases 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000035622 drinking Effects 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000008846 dynamic interplay Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000006996 mental state Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 208000037974 severe injury Diseases 0.000 description 1
- 230000009528 severe injury Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- 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
- G06F16/35—Clustering; Classification
-
- 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
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Primary Health Care (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Technology Law (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a legal three-level theory automatic reasoning method, a device and electronic equipment, wherein the method comprises the following steps: constructing a legal event evolution interaction map based on multiple rounds of interaction information; inputting the legal event evolution interaction map into a pre-training reasoning model to obtain a reasoning result of a legal event; the pre-training reasoning model is obtained by constructing proposition probabilistic representations generated by combining proposition representations of different knowledge types in the law event evolution interaction map by using a three-level theory. Because the predicted path is generated based on legal three-level theory reasoning, the prediction can have sufficient interpretability, for example, the condition that the requirement legal title is established and the probability size supporting the requirement can be explained clearly, and further prediction and evidence support are provided for legal decision.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to a legal three-level theory automatic reasoning method and device and electronic equipment.
Background
Inference is a very important and challenging task in the field of natural language processing, whose purpose is to make decisions on unseen input information using existing knowledge and inference techniques. Machine reasoning requires heuristics and strategies, which are usually done by knowledgeable domain experts. This process is where machine reasoning is difficult for an enterprise to scale, as it requires a lot of expert manpower to complete the strategy.
Machine reasoning is best suited for deterministic scenarios. That is, it is determined whether an event is real or whether it will occur. Machine Reasoning (Machine learning) has good application in tasks such as common sense question answering, fact detection, natural language Reasoning, visual common sense Reasoning, visual question answering, document level question answering, multiple rounds of semantic analysis and question answering.
For the specific field reasoning, the reasoning ability must be constructed in the specific field and the specific reasoning problem, and the technical and theoretical practices are integrated. For example, in the case of legal events, there is no complete knowledge base, and there may be deviation in the cognition of any expert, and the legal knowledge is dynamically updated, so it is difficult to obtain complete information. In addition, in addition to data such as support rate, success or failure rate, and the like, legal events have logic between events, so that it is difficult to realize a perfect reasoning result by simply using data and adopting machine reasoning to reason the legal events.
Disclosure of Invention
In order to solve the problem that the legal event reasoning in the prior art is difficult to obtain a perfect reasoning result, a legal three-level theory automatic reasoning method, a device and electronic equipment are provided.
In a first aspect, the invention provides an automatic reasoning method for a legal three-layer theory, which comprises the following steps: constructing a legal event evolution interaction map based on multiple rounds of interaction information; inputting the legal event evolution interaction map into a pre-training reasoning model to obtain a reasoning result of a legal event; the pre-training reasoning model is obtained by constructing proposition probabilistic representations generated by combining proposition representations of different knowledge types in the law event evolution interaction map by using a three-level theory.
Optionally, the constructing an event evolution interaction map based on multiple rounds of interaction information includes: acquiring a legal event map; acquiring a first requirement and a first requirement relation of a legal event described by an interactive object based on a multi-round interactive strategy; and updating the legal event map by using the first requirement and the first requirement relation to obtain a legal event evolution interaction map.
Optionally, the obtaining of the first requirement and the first requirement relationship of the legal event described by the interactive object based on the multiple rounds of interaction strategies includes: understanding a legal event described by an interactive object, and acquiring the field, intention, event and legal entity of the legal event; extracting the first requirement and the first requirement relation based on the field, intention, event and legal entity of the legal event; generating feedback information based on the legal event, wherein the feedback information is used for interacting with the interacted object; repeatedly executing the step to understand the legal events described by the interactive objects, and acquiring the field, intention, events and legal entities of the legal events; extracting the first requirement and the first requirement relation based on the field, intention, event and legal entity of the legal event; generating feedback information based on the legal events, wherein the legal event evolution interaction map can represent complete information of the legal events.
Optionally, the obtaining the legal event map comprises: defining a legal event map; collecting legal corpora on a legal event map; and extracting a second requirement and a second requirement relation from the legal corpus based on the legal corpus.
According to a second aspect, an embodiment of the present invention provides a method for building an automatic reasoning model of a legal three-level theory, including: constructing a legal event evolution interaction map based on multiple rounds of interaction information; performing legal proposition representation on the legal event graph by utilizing a reasoning representation framework corresponding to the knowledge type based on the knowledge type in the legal event evolution interaction graph; performing probabilistic representation on the legal proposition by using a three-level theory to obtain a legal proposition probability model; and constructing a reasoning model based on the legal propositional probability model.
Optionally, the inference representation framework comprises: at least one of a symbolic logic based representation framework, a statistical rule based representation framework, and an embedded vector representation framework.
Optionally, the representing the legal proposition probabilistically by using a three-layer theory includes: fusing legal propositions represented by different representation frames by utilizing a three-level theory; and 3, carrying out reasoning connection on the fused legal propositions by utilizing a probabilistic graph model.
Optionally, fusing legal propositions represented by different representation frames using a three-level theory includes: carrying out variation quantization on the legal propositions represented by different representation frames; and associating the variable legal propositions according to the evolution relation of the three-level theory.
Optionally, the constructing of the inference model based on the propositional probability model includes: gradually decomposing each proposition dependency relationship in the proposition probability graph to obtain the probability distribution of each legal proposition; and estimating model parameters based on the probability distribution of each legal proposition, and constructing the reasoning model.
According to a third aspect, an embodiment of the present invention provides an automatic reasoning apparatus for legal three-level theory, including: the first spectrum construction module is used for constructing a legal event evolution interaction spectrum based on multiple rounds of interaction information; the reasoning module is used for inputting the legal event evolution interaction map into a pre-training reasoning model to obtain a reasoning result of the legal event; the pre-training reasoning model is obtained by constructing proposition probabilistic representations generated by combining proposition representations of different knowledge types in the law event evolution interaction map by using a three-level theory.
According to a fourth aspect, an embodiment of the present invention provides an automatic reasoning model building apparatus for law three-level theory, including: the second spectrum construction module is used for constructing a legal event evolution interaction spectrum based on multiple rounds of interaction information; the proposition representation module is used for carrying out legal proposition representation on the legal event map by utilizing a reasoning representation framework corresponding to the knowledge type based on the knowledge type in the legal event evolution interaction map; the probability model building module is used for performing probabilistic representation on the legal proposition by utilizing a three-level theory to obtain a legal proposition probability model; and the reasoning model building module builds a reasoning model based on the law proposition probability model.
According to a fifth aspect, an embodiment of the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the law three-hierarchy automatic inference method according to any one of the first aspect and/or the law three-hierarchy automatic inference model construction method according to the second aspect
According to a sixth aspect, an embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the program to implement the automatic reasoning method according to any one of the first aspect and/or the automatic reasoning model building method according to any one of the second aspect.
Because the predicted path is generated based on legal three-level theory reasoning, the prediction can have sufficient interpretability, for example, the condition that the requirement legal title is established and the probability size supporting the requirement can be explained clearly, and further prediction and evidence support are provided for legal decision.
The method comprises the steps of constructing a legal event evolution interaction map based on multiple rounds of interaction information, serving static and incomplete features of the event map, adding interaction and evolution characteristics to the event map, continuously supplementing requirements and relations of a legal event in the process of interaction with a user and an expert, adjusting the current legal reasoning event, and finally obtaining complete relevant information so as to ensure that materials required by reasoning are accurate and complete.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a legal three-level automatic reasoning model construction method provided by an embodiment of the present invention;
FIG. 2 is a schematic diagram of a first round of event evolution interaction map provided by an embodiment of the present invention;
FIG. 3 is a schematic diagram of a second round of event evolution interaction map provided by the embodiment of the present invention;
FIG. 4 is a schematic diagram of an event evolution interaction map of a third round provided by the embodiment of the invention;
FIG. 5 is a schematic diagram of an event evolution interaction map of a fourth round according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of a multi-turn question-answering framework provided by an embodiment of the present invention;
FIG. 7 is a diagram of a probability of propositions in a three-level theory according to an embodiment of the present invention;
FIG. 8 is a flow chart illustrating an automatic reasoning method for legal three-layer theory according to an embodiment of the present invention;
fig. 9 is a block diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only partial embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and claims of this application and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It should be understood that the data so used may be interchanged under appropriate circumstances such that embodiments of the application described herein may be used. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.
As described in the background, the theoretical knowledge of machine reasoning is not complete, the philosophically executable reasoning is based on the human comprehension reasoning ability, the intelligence of the machine from the human is still a big gap, the reasoning ability must be built in a specific field and a specific reasoning problem, and the technology and theory are integrated into a practice. Most researches mainly depend on results output by a machine algorithm model, the interpretability of the results is very poor, and the results which are not interpreted according to the law cause and effect reasoning are difficult to be adopted and recognized. Fitting data simply through machine learning cannot generate interpretability, an entity and logic must be constructed under the guidance of theory, and the generated reasoning result can be interpreted through steps and paths of a reasoning theory.
The embodiment of the application provides a method for building an automatic reasoning model of a law three-level theory, which can comprise the following steps as shown in fig. 1:
and S11, constructing a legal event evolution interaction map based on multiple rounds of interaction information. Specifically, a question-answering engine and an event knowledge graph dynamic evolution combination method are adopted, and a first-layer knowledge representation and extraction method is provided for reasoning of a three-level theory. The legal event evolution interaction map does not depend on preset data, does not match a subgraph, and generates the subgraph. Compared with the prior art, the event evolution interaction map is not derived from a preset knowledge map, but is dynamically generated in the interaction process.
S12, performing legal proposition representation on the legal event map by utilizing a reasoning representation framework corresponding to the knowledge type based on the knowledge type in the legal event map;
and S13, performing probabilistic representation on the law proposition by using a three-level theory to obtain a probability model of the law proposition. The process of expressing the event evolution interaction map to a proposition probability map is that proposition recognizable by a computer is expressed by aiming at different legal propositions through a plurality of modes, different proposition modes are not regarded as independent decision information, but the theoretical knowledge of three-level theory is combined with a probability map model to carry out probability mapping on the proposition, and a second-level knowledge abstract expression method is provided. Compared with the prior art, the proposition representation of the knowledge graph usually depends on embedded learning, three types of learning based on symbols, statistical rules and embedded rules are fused, the proposition is combined into a probability graph by utilizing a three-level theory, and the decision information is further represented.
And S14, constructing a reasoning model based on the law proposition probability model.
Based on the learning and reasoning of the probabilistic graph model, a method using three-level graph relations as reasoning explanatory evidence is provided. Compared with the prior art, the training method of the model in the prior art can be adopted, and the three-level theory is used for explaining the result and the edge probability distribution in the embodiment, so that the model has the explaining capability, and can still execute reasoning even if the information of the segment is. The proposition with responsibility can also be inferred only by age and mental state, for example, so that the probability of the proposition supports the explanation basis.
Because the predicted path is generated based on legal three-level theory reasoning, the prediction can have sufficient interpretability, for example, the condition that the requirement legal title is established and the probability size supporting the requirement can be explained clearly, and further prediction and evidence support are provided for legal decision.
As an exemplary embodiment, a user interacts with an event map engine through a plurality of rounds of question and answer engines to gradually form an event evolution interaction map example aiming at legal intentions and legal scenes of the user, the example represents the stated complete information of the user and is represented in a knowledge map mode, and the recognition and the calculation are convenient for a computer. An exemplary event evolution interaction graph constructed by the dialog shown in fig. 2-5 is used as an example for illustration:
after the session starts, the user states: "i drink some liquor, drive and hit passerby, how do" identifies natural language based on question and answer engine, constructs the first round of event evolution interaction map shown in fig. 2: the behavior subject is 'my', criminal behavior 'driving after drinking' is implemented, and the harm result 'colliding' is caused. And based on the user's natural language and an action runner in the question-and-answer engine, such as a query for laws and regulations, understand the user's interactive content and generate corresponding dialog feedback to the user. Based on the user's interactive content, the following dialog may be generated: a machine: "ask for what the content of blood alcohol test is", user: "185 mg/ml", based on the natural language recognized by the question-answering engine, the event requirements and the relationship between the event requirements are supplemented on the basis of fig. 2, and a second round of event evolution interaction map shown in fig. 3 is constructed: the essential attributes of the add event are: the alcohol content was 185 mg/ml. Continuing to understand the user interaction context, a new dialog is generated: a machine: "ask what car you drive", user: "family Benz". Based on the question-answering engine to identify natural language, the event evolution interaction map of the third round shown in fig. 4 is constructed by supplementing the requirements of the events and the relationship between the requirements on the basis of fig. 3: the essential attributes of the add event are: the vehicle type is that of a galloping vehicle belonging to a common motor vehicle. Continuing to understand the user interaction context, a new dialog is generated: a machine: "hit several people, how well the hit passerby is", user: "hit 3, 1 severe injury 1 death". Based on the question-answering engine to identify natural language, the event evolution interaction map of the fourth round shown in fig. 5 is constructed by supplementing the requirements of the events and the relationship between the requirements on the basis of fig. 4: the requirements for the add event result is progressively: resulting in progressive crash and bruising. The four-round interactive question-answering in the embodiment is only an exemplary illustration of the construction process of the event evolution interaction map, and more rounds of interaction and fewer rounds of interaction are also within the protection scope of the embodiment of the present application.
And continuously identifying the supplemented essential elements such as behavior bodies, hazard behaviors, criminal objects, hazard results and the like and the relationship among the essential elements based on the event evolution interaction map, and finally, existing in the instance of the knowledge map.
The event evolution interaction map is constructed by a knowledge map generated by dynamic interaction updating, the information of the user is interacted with the user according to the intention of the user, the corresponding information of legal event scenes, elements, experiences and the like, the information of the user is obtained, the event evolution interaction map is dynamically constructed by system processing, and the map reveals complete information of legal propositions.
As an exemplary embodiment, a multi-turn question-and-answer framework for the identification process involved in the evolving interaction process is shown in FIG. 6, in particular, where
The XNLP represents a natural language processing module and is used for lexical analysis, syntactic analysis and the like of legal texts.
The XNLU represents a natural language understanding module for domain recognition, intent recognition, event recognition and legal entity recognition of legal text
The dialog Engine represents a dialog Engine, and comprises two core modules: state trackers and policy management. The state tracker records the context of the user interaction, and the policy management is used to distinguish the action returned by the current dialog machine.
The Action Runner represents an Action Runner that runs skills of the robot, such as inquiry of laws and regulations, and the like.
And the XNLG represents natural language generation, automatically generates a dialogue according to a feedback action, feeds the dialogue back to a user, feeds the currently collected information back to the event evolution interaction map, and updates the current map example.
And after the event evolution interaction map is generated, starting to enter a legal three-level reasoning proposition representation module. The hierarchy is a lower concept of crime, that is, a certain crime is composed of different conditions, and the conditions have a hierarchical relationship. The third level is that the crime is composed of three elements with a level relation. The hierarchy has dual meanings: the first layer means that the crime is a standing condition, i.e. the hierarchy is solid. The second level of meaning is the relationship between crime establishing requirements, i.e. the hierarchy is logical. The rank between crime establishing requirements means that the crime establishing requirements have the following two relations: regarding the relationship between the former and the latter, there is a relationship of "there is no latter, there is the former" as well; in the relationship between the latter requirement and the former requirement, there is a relationship of "if the former is not present, the latter is not present".
An act to constitute a crime must meet three conditions in a progressive combination.
(1) The crime constitutes the current. The crime construction compatibility is also called construction requirement conformity, and means the realization of construction requirements, that is, the fact that the occurrence is consistent with the content specified by the criminal law provisions. Specifically, the sex includes several elements of action subject, harm action, criminal object, harm result and cause and effect relationship.
(2) An violation. The law of law requires that criminal acts not only be compliant with the requirements of the constitution, but also be substantially impermissible by law, i.e., must be illegal. The criterion of the offensiveness is whether there is an offence barrier. An offence-prevention incident is an incident that precludes the offence of an act with that activity. The offending barriers generally include proper defense, emergency refuge, law acts, commitments from victims, and the like.
(3) Is responsible for this. Accountability means that the agent can be non-troubled and reprimmed with regard to the act of meeting the conditions of equality and violation. Whether or not there is liability should be examined in terms of criminal liability of the agent, deliberate or accidental crime, etc. In addition, there are two important reasons for liability, namely illegal recognition and lack of expectation.
In this embodiment, the knowledge type in the legal event graph may include the entity relationship of the essential element in the event, the attribute relationship of the essential element in the event, and descriptive text. Different inference representation frameworks are employed for different knowledge types in this embodiment, and the so-called inference representation frameworks may include a symbolic logic-based representation framework, a statistical rule-based representation framework, and an embedded vector representation framework. Therefore, in the embodiment, a plurality of reasoning representation modes are applied, and propositions are observed from different angles. Several representations are generated based on the specific knowledge of the event evolution interaction map, and since different knowledge fits different representation frameworks, multiple representations are required to be able to solve the problem. The factual entity relationships in the event that a common speed is a motor vehicle, for example, are consistent with a symbol-based logical representation, namely, the representation as < speed, is, motor vehicle >. The requirement attribute relationship of alcohol content and drunk driving can be statistically calculated from a large Number of cases, and forms knowledge, which is expressed as Number (alcohol content, quantity) > Action (drunk driving).
Some descriptive textual knowledge is difficult to represent in some formal language, which is a solution using representation learning, and representing propositions as high-dimensional vectors, e.g. "breathe while alighting" is difficult to directly map to propositions S (death or death), which can be represented as high-dimensional vectors in this embodiment based on a large number of pre-trained corpora, i.e. S ("breathe while alighting") [1.2,2.34,63.4, …,5.6 ]. The expression based on symbolic logic refers to extracting corresponding ontology content from the event evolution interaction map, and reasoning the relationship between concept classification and concept based on the rule of a production formula. The event interaction evolution map representation can be expressed as: symbolic logic, statistical rules and embedded vectors, and the three data completely store fact requirement information.
The three-level theory is used for representing the law proposition in a probabilistic manner to obtain a law proposition probability model, specifically, the three-level theory proposition probability graph shown in figure 7 can be used for representing the law proposition in a probabilistic manner, the event evolution interaction graph comprises a large amount of event knowledge, the event knowledge is required to serve the three-level theory, the facts and the proposition need to be quantized, after the quantization, the probability graph model is used for associating according to the evolution relation of the three-level theory, and finally the representing process is finished
Specifically, referring to the probability diagram of the proposition in the three-level theory shown in fig. 7, propositions are represented by x1, x2, x3, …, xn: an action subject proposition of x1, an attack object proposition of x2, an action result establishment proposition of x3, a causal relationship proposition of x4, an intentional proposition of x5, an intention proposition of x5, an objective constituent proposition of x6, a subjective constituent proposition of x7, a constituent proposition of legitimacy of x8, and the like. The overall probability formula is as follows:
the method comprises the following steps of decomposing step by step according to the dependency relationship in the graph, converting corresponding marginal probability distribution in case data, and further performing overall distribution construction, such as objectively constructing requirement propositions:
P(x6|x1，x2，x3）
the key element legality distribution:
P(x8|x6，x7)
after each distribution is obtained, model learning and reasoning can be performed.
If it is assumed that the crime is true proposition Y, then parameter estimation can be done according to the following formula:
wherein, Y is reasoning model parameter, Y is crime establishment proposition, and X is law proposition.
After the model parameters are estimated, the construction of a reasoning model is completed, and when a new event interaction evolution diagram exists, the current model can be used for prediction. And because the predicted path is generated based on the law three-level theory reasoning, the prediction can have sufficient interpretability, for example, the condition that the requirement legal title is established and the probability size supporting the condition can be explained clearly, and further prediction and evidence support are provided for legal decision.
The embodiment of the present application further provides an automatic inference method for law three-level theory, as shown in fig. 8, the method may include the following steps:
and S21, constructing a legal event evolution interaction map based on multiple rounds of interaction information. Reference may be made in detail to the description of the construction of the legal event evolution interaction map based on multiple rounds of interaction information in the above embodiments.
And S22, inputting the law event evolution interaction map into a pre-training reasoning model to obtain a reasoning result of the law event. The inference model is specifically obtained by constructing proposition probabilistic representations generated by combining proposition representations of different knowledge types in the legal event evolution interaction map by using a three-level theory. See the method for reasoning model construction in the above embodiments.
Because the predicted path is generated based on legal three-level theory reasoning, the prediction can have sufficient interpretability, for example, the condition that the requirement legal title is established and the probability size supporting the requirement can be explained clearly, and further prediction and evidence support are provided for legal decision.
Constructing the legal event evolution interaction map based on the multiple rounds of interaction information may include: acquiring a legal event map; acquiring a first requirement and a first requirement relation of a legal event described by an interactive object based on a multi-round interactive strategy; and updating the legal event map by using the first requirement and the first requirement relation to obtain a legal event evolution interaction map. Specifically, the legal time map can be obtained by defining a legal event map; collecting legal corpora on a legal event map; and extracting a second requirement and a second requirement relation from the legal corpus based on the legal corpus. The first requirement may be a requirement of a legal event described by a user, which is obtained in an interaction process, and the second requirement may be identified in a legal corpus based on the legal corpus, such as a legal document, a legal volume, and the like.
The method comprises the steps of constructing a legal event evolution interaction map based on multiple rounds of interaction information, serving static and incomplete features of the event map, adding interaction and evolution characteristics to the event map, continuously supplementing requirements and relations of a legal event in the process of interaction with a user and an expert, adjusting the current legal reasoning event, and finally obtaining complete relevant information so as to ensure that materials required by reasoning are accurate and complete.
An embodiment of the present invention provides an electronic device, as shown in fig. 9, the electronic device includes one or more processors 31 and a memory 32, and one processor 33 is taken as an example in fig. 9.
The controller may further include: an input device 33 and an output device 34.
The processor 31, the memory 32, the input device 33 and the output device 34 may be connected by a bus or other means, and the bus connection is exemplified in fig. 9.
The processor 31 may be a Central Processing Unit (CPU). The processor 31 may also be other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or combinations thereof. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory 32, which is a non-transitory computer readable storage medium, can be used for storing non-transitory software programs, non-transitory computer executable programs, and modules, such as program instructions/modules corresponding to the control methods in the embodiments of the present application. The processor 31 executes various functional applications and data processing of the server by running non-transitory software programs, instructions and modules stored in the memory 32, that is, implementing the law three-hierarchy automatic reasoning method and/or the law three-hierarchy automatic reasoning model building method of the above-described method embodiments.
The memory 32 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to use of a processing device operated by the server, and the like. Further, the memory 32 may include high speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory 32 may optionally include memory located remotely from the processor 31, which may be connected to a network connection device via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input device 33 may receive input numeric or character information and generate key signal inputs related to user settings and function control of the processing device of the server. The output device 34 may include a display device such as a display screen.
One or more modules are stored in the memory 32, and when executed by the one or more processors 31, perform the methods as shown in fig. 1 and/or 8.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program to instruct related hardware, and the program can be stored in a computer readable storage medium, and when executed, the program can include the processes of the embodiments of the motor control methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-only memory (ROM), a Random Access Memory (RAM), a flash memory (FlashMemory), a hard disk (hard disk drive, abbreviated as HDD) or a Solid State Drive (SSD), etc.; the storage medium may also comprise a combination of memories of the kind described above.
Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations fall within the scope defined by the appended claims.
Finally, the principle and the implementation of the present invention are explained by applying the specific embodiments in the present invention, and the above description of the embodiments is only used to help understanding the method and the core idea of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
After the model parameters are estimated, when a new event interaction evolution diagram exists, the current model can be used for prediction. And because the predicted path is generated based on the law three-level theory reasoning, the prediction can have sufficient interpretability, for example, the condition that the requirement legal title is established and the probability size supporting the condition can be explained clearly, and further prediction and evidence support are provided for legal decision.
Claims (10)
1. An automatic reasoning method for a law three-layer theory is characterized by comprising the following steps:
constructing a legal event evolution interaction map based on multiple rounds of interaction information;
inputting the legal event evolution interaction map into a pre-training reasoning model to obtain a reasoning result of a legal event;
the pre-training reasoning model is obtained by constructing proposition probabilistic representations generated by combining proposition representations of different knowledge types in the law event evolution interaction map by using a three-level theory.
2. The legal three-level theory automatic reasoning method of claim 1, wherein the building of the event evolution interaction graph based on multiple rounds of interaction information comprises:
acquiring a legal event map;
acquiring a first requirement and a first requirement relation of a legal event described by an interactive object based on a multi-round interactive strategy;
and updating the legal event map by using the first requirement and the first requirement relation to obtain a legal event evolution interaction map.
3. The legal three-level theory automatic reasoning method of claim 2, wherein acquiring the first requirement and the first requirement relationship of the legal event described by the interactive object based on the multiple rounds of interaction strategies comprises:
understanding a legal event described by an interactive object, and acquiring the field, intention, event and legal entity of the legal event;
extracting the first requirement and the first requirement relation based on the field, intention, event and legal entity of the legal event;
generating feedback information based on the legal event, wherein the feedback information is used for interacting with the interacted object;
repeatedly executing the step to understand the legal events described by the interactive objects, and acquiring the field, intention, events and legal entities of the legal events; extracting the first requirement and the first requirement relation based on the field, intention, event and legal entity of the legal event; generating feedback information based on the legal events, wherein the legal event evolution interaction map can represent complete information of the legal events.
4. The legal three-layer theory automatic reasoning method of claim 2, wherein the obtaining the legal event graph comprises:
defining a legal event map;
collecting legal corpora on a legal event map;
and extracting a second requirement and a second requirement relation from the legal corpus based on the legal corpus.
5. A method for constructing an automatic reasoning model of a law three-layer theory is characterized by comprising the following steps:
constructing a legal event evolution interaction map based on multiple rounds of interaction information;
performing legal proposition representation on the legal event graph by utilizing a reasoning representation framework corresponding to the knowledge type based on the knowledge type in the legal event evolution interaction graph;
performing probabilistic representation on the legal proposition by using a three-level theory to obtain a legal proposition probability model;
and constructing a reasoning model based on the legal propositional probability model.
6. The legal three-level automatic reasoning model building method of claim 5, wherein the reasoning representation framework comprises: at least one of a symbolic logic based representation framework, a statistical rule based representation framework, and an embedded vector representation framework.
7. The method of claim 1, wherein the probabilistic representation of the legal proposition using a three-level theory comprises:
fusing legal propositions represented by different representation frames by utilizing a three-level theory;
and 3, carrying out reasoning connection on the fused legal propositions by utilizing a probabilistic graph model.
8. The method of claim 7, wherein fusing legal propositions represented by different representation frames using a three-level theory comprises:
carrying out variation quantization on the legal propositions represented by different representation frames;
and associating the variable legal propositions according to the evolution relation of the three-level theory.
9. The legal three-level automatic reasoning model building method of claim 5, wherein the building of the reasoning model based on the propositional probability model comprises:
gradually decomposing each proposition dependency relationship in the proposition probability graph to obtain the probability distribution of each legal proposition;
and estimating model parameters based on the probability distribution of each legal proposition, and constructing the reasoning model.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of legal three-level theory automatic inference as claimed in any one of claims 1-4 and/or the method of legal three-level theory automatic inference model construction as claimed in any one of claims 5-9 when executing the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010532799.4A CN111797230B (en) | 2020-06-11 | 2020-06-11 | Legal three-layer theory automatic reasoning method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010532799.4A CN111797230B (en) | 2020-06-11 | 2020-06-11 | Legal three-layer theory automatic reasoning method and device and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111797230A true CN111797230A (en) | 2020-10-20 |
| CN111797230B CN111797230B (en) | 2021-07-13 |
Family
ID=72804324
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010532799.4A Active CN111797230B (en) | 2020-06-11 | 2020-06-11 | Legal three-layer theory automatic reasoning method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111797230B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115358896A (en) * | 2022-10-20 | 2022-11-18 | 四川大学华西医院 | Method, device, equipment and medium for constructing criminal evolution network by using massive documents |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070288272A1 (en) * | 2006-05-17 | 2007-12-13 | Marks Peter T | Collection systems and methods for managing insurance subrogation claims |
| CN108304386A (en) * | 2018-03-05 | 2018-07-20 | 上海思贤信息技术股份有限公司 | A kind of logic-based rule infers the method and device of legal documents court verdict |
| US20180330456A1 (en) * | 2017-05-12 | 2018-11-15 | Bayo Odutola | System and method for detecting, profiling and benchmarking intellectual property professional practices and the liability risks associated therewith |
| CN109377437A (en) * | 2018-11-21 | 2019-02-22 | 南京莱斯信息技术股份有限公司 | Urban Emergency emergency inference pattern based on event chain |
| CN109741002A (en) * | 2018-11-28 | 2019-05-10 | 广州法通信息咨询服务有限公司 | A kind of procuratorate remotely appeals system |
| CN109977237A (en) * | 2019-05-27 | 2019-07-05 | 南京擎盾信息科技有限公司 | A kind of dynamic law occurrence diagram spectrum construction method towards legal field |
| CN110046262A (en) * | 2019-06-10 | 2019-07-23 | 南京擎盾信息科技有限公司 | A kind of Context Reasoning method based on law expert's knowledge base |
| CN110059193A (en) * | 2019-06-21 | 2019-07-26 | 南京擎盾信息科技有限公司 | Legal advice system based on law semanteme part and document big data statistical analysis |
| CN110232447A (en) * | 2019-04-28 | 2019-09-13 | 杭州实在智能科技有限公司 | Legal case depth reasoning method |
| CN110377759A (en) * | 2019-07-22 | 2019-10-25 | 中国工商银行股份有限公司 | Event relation map construction method and device |
-
2020
- 2020-06-11 CN CN202010532799.4A patent/CN111797230B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070288272A1 (en) * | 2006-05-17 | 2007-12-13 | Marks Peter T | Collection systems and methods for managing insurance subrogation claims |
| US20180330456A1 (en) * | 2017-05-12 | 2018-11-15 | Bayo Odutola | System and method for detecting, profiling and benchmarking intellectual property professional practices and the liability risks associated therewith |
| CN108304386A (en) * | 2018-03-05 | 2018-07-20 | 上海思贤信息技术股份有限公司 | A kind of logic-based rule infers the method and device of legal documents court verdict |
| CN109377437A (en) * | 2018-11-21 | 2019-02-22 | 南京莱斯信息技术股份有限公司 | Urban Emergency emergency inference pattern based on event chain |
| CN109741002A (en) * | 2018-11-28 | 2019-05-10 | 广州法通信息咨询服务有限公司 | A kind of procuratorate remotely appeals system |
| CN110232447A (en) * | 2019-04-28 | 2019-09-13 | 杭州实在智能科技有限公司 | Legal case depth reasoning method |
| CN109977237A (en) * | 2019-05-27 | 2019-07-05 | 南京擎盾信息科技有限公司 | A kind of dynamic law occurrence diagram spectrum construction method towards legal field |
| CN110046262A (en) * | 2019-06-10 | 2019-07-23 | 南京擎盾信息科技有限公司 | A kind of Context Reasoning method based on law expert's knowledge base |
| CN110059193A (en) * | 2019-06-21 | 2019-07-26 | 南京擎盾信息科技有限公司 | Legal advice system based on law semanteme part and document big data statistical analysis |
| CN110377759A (en) * | 2019-07-22 | 2019-10-25 | 中国工商银行股份有限公司 | Event relation map construction method and device |
Non-Patent Citations (1)
| Title |
|---|
| 杨雪蓉等: "事件关系检测的语言学资源、评测及研究综述", 《智能计算机与应用》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115358896A (en) * | 2022-10-20 | 2022-11-18 | 四川大学华西医院 | Method, device, equipment and medium for constructing criminal evolution network by using massive documents |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111797230B (en) | 2021-07-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Walton | Methods of argumentation | |
| Horsten et al. | Truth is simple | |
| Magnani | The eco-cognitive model of abduction II: Irrelevance and implausibility exculpated | |
| Cariani et al. | Conditionals, context, and the suppression effect | |
| Walton | Some artificial intelligence tools for argument evaluation: An introduction | |
| CN111797230B (en) | Legal three-layer theory automatic reasoning method and device and electronic equipment | |
| Gabbay et al. | The practical turn in logic | |
| Engelmann et al. | Argumentation as a method for explainable AI: A systematic literature review | |
| CN111797232B (en) | Legal event map construction method, legal event reasoning method, device and electronic equipment | |
| Galitsky | Exhaustive simulation of consecutive mental states of human agents | |
| Schneiderhan | Genocide reconsidered: A pragmatist approach | |
| CN111797231B (en) | Three-level theory-based legal reasoning proposition representation method, legal event reasoning method and electronic equipment | |
| CN115982335A (en) | Active AI medical question-answering system, method, equipment and storage medium | |
| Hanson et al. | Introduction: Return of the A Priori | |
| Walton | How to refute an argument using artificial intelligence | |
| Oaksford et al. | The uncertain reasoner: Bayes, logic, and rationality | |
| Lawley et al. | Towards natural language story understanding with rich logical schemas | |
| Nyamsuren et al. | Human reasoning module | |
| CN111897959A (en) | Method, apparatus, device and storage medium for reasoning within dynamic legal events | |
| Gryz et al. | Futility of a Right to Explanation. | |
| Gueddes et al. | Enhancing ontology-based home Care Services platform using Bayesian networks | |
| Kamateri et al. | Knowledge-based intelligence and strategy learning for personalised virtual assistance in the healthcare domain | |
| Duplessis et al. | A conventional dialogue model based on empirically specified dialogue games | |
| Galitsky et al. | Managing customer relations in an explainable way | |
| Keeley Crockett et al. | Bullying and debt: Developing novel applications of dialogue systems |
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 |