CN110347814B - Lawyer accurate recommendation method and system - Google Patents

Lawyer accurate recommendation method and system Download PDF

Info

Publication number
CN110347814B
CN110347814B CN201910573715.9A CN201910573715A CN110347814B CN 110347814 B CN110347814 B CN 110347814B CN 201910573715 A CN201910573715 A CN 201910573715A CN 110347814 B CN110347814 B CN 110347814B
Authority
CN
China
Prior art keywords
lawyer
case
score
user
lawyers
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910573715.9A
Other languages
Chinese (zh)
Other versions
CN110347814A (en
Inventor
来悦
李建元
彭俊江
陈涛
张迅
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Yinjiang Technology Co.,Ltd.
Original Assignee
Enjoyor Co Ltd
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 Enjoyor Co Ltd filed Critical Enjoyor Co Ltd
Priority to CN201910573715.9A priority Critical patent/CN110347814B/en
Publication of CN110347814A publication Critical patent/CN110347814A/en
Application granted granted Critical
Publication of CN110347814B publication Critical patent/CN110347814B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G06F16/33—Querying
    • G06F16/332—Query formulation
    • G06F16/3329—Natural language query formulation or dialogue systems
    • 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)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Tourism & Hospitality (AREA)
  • Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Technology Law (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • General Business, Economics & Management (AREA)
  • Human Computer Interaction (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

A lawyer accurate recommendation method includes the steps of firstly obtaining case description input by a user, then obtaining key words through word segmentation, and obtaining similarity between a case and user problems through calculating similarity between the case and expansion words of the case and each key word. And then obtaining professional matching degree scores of lawyers according to case information transacted by the lawyers in the database. And then calculating to obtain four evaluation items of lawyer practice period, case handling duration, lawyer winning rate and expert score, and endowing corresponding weight to obtain professional ability score of the lawyer. And then, according to the location of the user, obtaining two evaluation items, namely the distance between the lawyer and the user and the number of cases being accepted by the lawyer, and giving corresponding weight to obtain the service convenience score of the lawyer. And finally, integrating the three dimensions to obtain the comprehensive evaluation score of the lawyer, and recommending the lawyer to the user. And a lawyer accurate recommendation system. The invention has high accuracy and better service convenience.

Description

Lawyer accurate recommendation method and system
Technical Field
The invention relates to the technical field of data processing, in particular to a lawyer accurate recommendation method and system.
Background
By the end of 2018, the whole country has 42.3 thousands of medical lawyers, and the number of the lawyers is more than 3 thousands. Nowadays, the lawyer industry in China plays more and more important roles in promoting the development of economic society, maintaining the rights of people and guaranteeing the fair sense of society. And for the public, how to find out lawyers suitable for the public from more than 40 million lawyers to successfully maintain the legal rights of the public and put forward the appeal of the public becomes a difficult problem. In order to solve the problem, it is necessary to integrate the existing lawyers and case related information through an information means, evaluate the lawyers by integrating multiple dimensions, and recommend the corresponding lawyers in combination with the user appeal.
The prior lawyer recommendation method comprises the following steps: patent CN201710759833.X 'A attorney recommendation method and system' provides an attorney recommendation method, which comprises the steps of firstly obtaining the similarity of a target case and cases in judge documents in a database based on a neighbor algorithm, and then determining recommended attorneys from the judge documents with high similarity according to the attorney winning rate and the law winning rate. Although the method considers the proficiency field of lawyers, the recommendation accuracy is not high because the professional ability of the lawyers is considered from one dimension of the winning rate. Patent CN201810271936.6, "a lawyer evaluation method and recommendation based on big data", comprehensively evaluates the professional abilities of lawyers from five aspects of similarity between lawyers 'responses and vocabularies in professional areas, cited legal provision rate, text length, poor response rate and response similarity, and makes lawyer recommendations in combination with lawyers' proficiency in professional areas. Although lawyers with high professional level can be recommended to a user more accurately by the method, the idleness degree of the lawyers and the geographic positions of the lawyers are not considered, the situation that a plurality of cases are simultaneously handled on the lawyers is recommended, and the problem that closed service loops cannot be solved exists.
Disclosure of Invention
In order to overcome the limitation of the conventional lawyer recommendation method, the invention provides the lawyer accurate recommendation method and system with high recommendation accuracy, which can comprehensively evaluate the lawyer in the adequacy field, professional ability and service convenience.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a lawyer accurate recommendation method comprises the following steps:
acquiring case description input by a user, and performing word segmentation processing to acquire key words;
calculating the similarity between the case and the expansion vocabulary thereof and the key vocabulary, and acquiring the similarity between the case and the description of the case;
calculating professional matching degree scores of each lawyer under the case description input by the user based on the similarity between the case description and the case description, the total number of the cases transacted by each lawyer and the number of the cases transacted by each lawyer under each case description;
the interface displays the attorney recommendation to the user with the top score according to the ranking of the scores.
Further, the word segmentation process obtains a key vocabulary, and the process is as follows:
adding a judicial vocabulary into a custom dictionary of a word segmentation tool, and segmenting the text described by the case to obtain segmented words; further, the noun, the verb and the first name verb are reserved, the repeated vocabulary is deleted, and the key vocabulary is obtained.
Further, the similarity between the acquired case and the description of the case is obtained by the following process:
let W1For a case in a database and its extended vocabulary set, W2Describing a set of key words for a user case, W1＝{w11,w12,…,w1eWherein e is W1Number of words contained, w11,w12,w1eAre respectively W11, 2, e of (1), W2＝{w21,w22,…,w2fWherein f is W2Number of key words involved, w21,w22,w2fAre respectively W21, 2 and f vocabularies in the specification;
W1and W2Similarity is as
Figure GDA0003145365440000021
Wherein Sim (w)1h，w2g) (h 1, 2., e, g 1, 2., f) obtained by a semantic model;
then, the similarity Sim (W) is calculated1，W2) And (6) carrying out normalization processing.
Further, the case and the extended vocabulary set thereof are obtained by the following method:
a large number of various types of judicial texts and social news are obtained as linguistic data, a semantic model is trained by using a language model, case-based entries are expanded by combining the judicial entries and the semantic model, and the case-based entries and the extended vocabulary sets are correspondingly stored in a database to form a case and an extended vocabulary set thereof.
Further, calculating the professional matching degree score of each lawyer under the case description input by the user, wherein the process is as follows:
assuming a total of n lawyers and m cases, the calculation expression is as follows:
Figure GDA0003145365440000031
wherein xijNumber of cases belonging to case j, s, representing lawyer i transactsjSimilarity of case description of user by j, NiRepresents the total number of cases, Z, of bar iiRepresents the professional matching degree score of lawyer i, wherein i is more than or equal to 1 and less than or equal to n and j is more than or equal to 1 and less than or equal to m.
Further, the lawyer recommendation method further comprises the following steps: integrating lawyer professional matching degree scores, lawyer professional ability scores and lawyer service convenience scores, calculating to obtain lawyer comprehensive scores, sorting according to the comprehensive scores, and displaying the lawyers with the scores higher in the front on an interface to recommend the lawyers to the user; wherein the lawyer professional ability score evaluation indexes comprise lawyer practice period, case handling duration, lawyer victory rate and expert score; the bar service convenience score evaluation index comprises the number of cases being accepted by the bar and the distance between the bar and the geographic position of the user.
Further, the lawyer professional ability score is calculated by the following method:
using logarithmic function to quantify the relationship between the law enforcement period and the score, and establishing an evaluation item F of the law enforcement period1:
F1＝ln(tz+2)
In the formula, tzLaw enforcement years, tz+2 to avoid the case of attorneys with a practice age of one year and less than one year with a score of zero for the evaluation item;
calculating the handling time of lawyers for handling similar cases of the user, and establishing an evaluation item F of the' case handling time2：
Figure GDA0003145365440000032
Wherein m is the number, xjTotal number of cases belonging to case j, t, handled by lawyersskFor the case of the kth case in j,sjthe similarity of case description information of case j and the user is set;
calculating the winning rate of the lawyer handling the same type case of the user and establishing an evaluation item F of the lawyer winning rate3：
Figure GDA0003145365440000041
In the formula, x0jTotal number of cases submitted by lawyers belonging to the victory of case j;
introducing an expert rating system to establish an evaluation item F of' expert rating4Extracting a plurality of lawyers at intervals to score experts, and scoring the lawyers by the experts according to the credit conditions of the lawyers and the performances in specific cases to finally obtain a comprehensive score;
integrating the 4 indexes to calculate the professional ability score of the lawyer;
Fi＝[F1 F2 F3 F4][λ1 λ2 λ3 λ4]T
in the formula, FiIs the professional competence score of lawyer i, [ F1 F2 F3 F4]Is a score vector, [ lambda ]1 λ2 λ3 λ4] TIs the transpose of the weight vector.
Further, the lawyer service convenience score is calculated by the following method:
establishing an evaluation item B of' number of cases being accepted1:
Figure GDA0003145365440000042
Wherein p is the number of cases being processed by the lawyer;
establishing an evaluation item B of' lawyer and user far and near2:
Figure GDA0003145365440000043
Then, integrating the 2 indexes to calculate the service convenience score of the lawyer;
Bi＝[B1 B2][μ1 μ2]T
in the formula, BiScore for convenience of service for lawyer i, [ B1 B2]To score the vector, [ mu ]1 μ2]TIs the transpose of the weight vector.
Further, the calculation of lawyer comprehensive score can input the weighted values of three indexes of lawyer professional matching degree, lawyer professional ability and service convenience according to the user requirement, and the final score is as follows:
Figure GDA0003145365440000044
in the formula, ScoreiIs the composite score of lawyer i, [ Zi Fi Bi]In order to be a score vector, the score vector,
Figure GDA0003145365440000051
is the transpose of the weight vector.
An attorney recommendation system, comprising:
a database module: the system is used for storing lawyer basic information, case bases and an extended vocabulary set thereof;
the basic data processing module: comprises a lawyer information processing unit and a case and vocabulary expansion unit;
the lawyer information processing unit is used for correlating basic information of lawyers with basic information of cases, and calculating the working years of the lawyers, the working duration of the cases and the number of the cases being worked by the lawyers;
the case routing vocabulary extension unit is used for updating and acquiring case routing extension vocabularies;
the man-machine interaction module: the system comprises a user input unit and a lawyer recommendation display unit;
the user input unit is used for providing a lawyer recommendation index which is input by a user for case description, positioning and selecting, wherein the lawyer recommendation index can be selected from lawyer professional matching degree or lawyer comprehensive index; the lawyer comprehensive indexes comprise lawyer professional matching degree, lawyer professional ability and lawyer service convenience, and a user can input the weight values of the three indexes according to the requirement;
the lawyer recommendation display unit is used for displaying lawyers with the scores close to the front, and a user can check the representation of the lawyer by clicking the lawyer and select the lawyer of the heart instrument; the lawyer portrait comprises a lawyer static portrait and a lawyer dynamic portrait, wherein the lawyer static portrait comprises a lawyer name, a gender, a practice license number, a law contact information and the like; the bar chart comprises bar chart parameters, bar chart parameters and the like.
Lawyer recommendation module: and the system is used for calculating lawyer professional matching degree scores or lawyer comprehensive scores according to user selection and sorting based on the case situation description.
The invention has the following beneficial effects:
1. lawyer professional field recommendation accuracy is high. Firstly, the patent adopts a self-defined judicial vocabulary to cut words, and can reduce the situation of wrong segmentation of legal professional vocabularies. Secondly, a large number of various types of judicial texts and social news are used as linguistic data to train a semantic model, so that the result is more accurate.
2. Lawyer professional ability recommendation accuracy is high. Because different lawyers have different degrees of excellence in cases in different fields, the similarity is introduced when the professional ability of the lawyer is evaluated, and therefore the professional ability level of the lawyer in the field concerned by the user can be obtained.
3. Comprehensively considering the service convenience of lawyers. The conventional lawyer recommendation cannot display the number of cases being transacted by a lawyer, and the method also incorporates the index into a lawyer recommendation system, so that the recommendation result is more reliable.
Drawings
Fig. 1 is a flow chart of a lawyer precision recommendation method.
Fig. 2 is a functional block diagram of a lawyer precision recommendation system.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
Referring to fig. 1, a lawyer accurate recommendation method includes the following steps:
1. obtaining base data
All data in this example are sourced from a city bureau. Acquiring basic lawyer information and basic case information and storing the acquired basic lawyer information and basic case information in a database. The basic information of lawyer includes lawyer number, lawyer name, lawyer gender, lawyer age, lawyer place, medical practice license number, time of first medical practice, contact information and the like. The case basic information comprises case numbers, case reasons, case acceptance time, case settlement time, case states, case results, acceptance attorney numbers and the like.
2. Processing underlying data
And (4) associating the basic lawyer information with the basic case information through the acceptance lawyer number to form a new table. The working period of the lawyer (unit is year), the case handling period (unit is month) and the number of cases being handled by the lawyer are calculated.
A large number of various types of judicial texts and social news are used as linguistic data, and a semantic model is trained by using a language model. In this embodiment, a Word2Vec model is used to train a skip-gram model, and then the similarity between two words can be calculated by using the model. And then expanding the schema in the database by combining the semantic model and the judicial vocabulary entry to form an expanded vocabulary which is stored in the database.
3. Obtaining key words of user problems
And acquiring case description input by a user, and performing word segmentation processing to acquire key words. The method specifically comprises the following steps: adding a judicial vocabulary into a custom dictionary of a word segmentation tool, and segmenting the text described by the case to obtain segmented words; further, the noun, the verb and the first name verb are reserved, the repeated vocabulary is deleted, and the key vocabulary is obtained.
In this example, suppose that the user inputs "husband and wife are away from the child and nurses the child with their mother, the father uses the obtained property to pay for nurturing the child at one time, the child is 18 years old and enters university, and the mother proposes to pay for the father to pay for part of the university of the child. Asking for a question: how is this situation legally stipulated? The method comprises the steps of performing word segmentation processing on the text, further keeping nouns, verbs and number of initials, and deleting repeated words to obtain key words. The key vocabulary obtained is: a couple, a dissociate, a child, a group, a mother, a foster, a father, a point, a property, a balance, a foster fee, a college, a mother, a proposal, an important, a burden, a part, a cost, a request, a condition, a law, a regulation.
4. Similarity calculation
And calculating the similarity between the case and the extended vocabulary and the key vocabulary thereof, and acquiring the similarity between the case and the description of the case. The method specifically comprises the following steps:
let W1For a case in a database and its extended vocabulary set, W2Describing a set of key words for a user case, W1＝{w11,w12,…,w1eWherein e is W1Number of words contained, w11,w12,w1eAre respectively W11, 2, e of (1), W2＝{w21,w22,…,w2fWherein f is W2Number of key words involved, w21,w22,w2fAre respectively W21, 2 and f vocabularies in the specification;
W1and W2Similarity is as
Figure GDA0003145365440000071
Wherein Sim (w)1h，w2g) (h 1, 2., e, g 1, 2., f) obtained by a semantic model;
then, the similarity Sim (W) is calculated1，W2) And (6) carrying out normalization processing.
In this example W2{ 'couple', 'dissociate', 'child', 'home', 'mother', 'foster', 'father', 'score', 'property', 'counter', 'foster', 'test up', 'down', 'up', 'down', 'up', and the like', ' university ', ' mother ', ' propose ', ' charge ', ' part ', ' charge ', ' ask ', ' situation ', ' law ', ' specify ' }
Similarity is as
Figure GDA0003145365440000081
Wherein Sim (w)1h，w2g) And (h 1, 2., e, g 1, 2., f) is obtained by a skip-gram model in Word2 vec. Then, each similarity is squared to enlarge the difference, and then divided by the maximum value to perform normalization processing.
5. Obtaining lawyer professional matching degree score
And acquiring the total number of the cases handled by each lawyer and the number of the cases handled by each lawyer under each case, and calculating the professional matching degree score of each lawyer under the case description of a certain user according to the similarity between the case and the case description. Assuming a total of n lawyers and m cases, the calculation expression is as follows:
Figure GDA0003145365440000082
in the formula xijNumber of cases belonging to case j, s, representing lawyer i transactsjSimilarity of case description of user by j, NiRepresents the total number of cases, Z, of bar iiRepresents the professional matching degree score of lawyer i, wherein i is more than or equal to 1 and less than or equal to n and j is more than or equal to 1 and less than or equal to m.
In this embodiment, according to the case of a lawyer, the matching degree score of the lawyer professional obtained through the above calculation is 0.111078.
And calculating all lawyers in the database to obtain professional matching degree scores of all lawyers. For descending ranking of scores, table 1 shows the attorneys who scored 10 times better for professional match in this case.
Figure GDA0003145365440000083
Figure GDA0003145365440000091
TABLE 1
As another embodiment, a lawyer recommendation method further comprises: integrating lawyer professional matching degree scores, lawyer professional ability scores and lawyer service convenience scores, calculating to obtain lawyer comprehensive scores, sorting according to the comprehensive scores, and displaying the lawyers with the scores higher in the front on an interface to recommend the lawyers to the user; wherein the lawyer professional ability score evaluation indexes comprise lawyer practice period, case handling duration, lawyer victory rate and expert score; the bar service convenience score evaluation index comprises the number of cases being accepted by the bar and the distance between the bar and the geographic position of the user.
6. Obtaining lawyer professional ability scores
The indexes for measuring the professional ability of lawyers comprise the practice period, the case acceptance duration, the lawyer winning rate and the expert score. Generally, the longer the lawyer practice years, the more experienced; while attorneys with higher practice years should not score too high in order to leave younger attorneys a chance to be recommended. Therefore, the relationship between the law enforcement period and the score can be quantified by using a logarithmic function, and an evaluation term F of the law enforcement period can be established1:
F1＝ln(tz+2)
In the formula, tzLaw enforcement years, tz+2 to avoid the case that the evaluation item score is zero for lawyers with a practice period of one year or less, normalization is performed.
In this embodiment, the evaluation item F of lawyer "lawyer practice period1As shown in table 2 (ellipses are not shown, in descending order).
Figure GDA0003145365440000092
Figure GDA0003145365440000101
TABLE 2
The transaction duration of different types of cases is greatly different, the similarity between the case description and the case description can be used to calculate the transaction duration of lawyer processing users of the same type of cases, and an evaluation item F of 'case transaction duration' is established2:
Figure GDA0003145365440000102
Wherein m is the number, xjTotal number of cases belonging to case j, t, handled by lawyersskThe handling time (in month) of the kth case in the case of j, sjAnd j is the similarity of case description and case description of the user.
In this embodiment, the evaluation item F of lawyer "duration of case handling2As shown in table 3 (ellipses are not shown, in descending order).
Lawyer number Lawyer name F2
615 Wu (Wu xi) 1.000000
1933 Liu (PZQ DXW) 1.000000
639 Comment 0.996855
393 Xu xi 0.989509
361 Wu (PZQ DXW) 0.933892
1633 Xu (PZQ DXW) 0.918576
161 Shen (Chinese character of 'Shen') 0.843257
1973 Lee (PZQ DXW) 0.835115
2078 Shen (Chinese character of 'Shen') 0.714286
1771 Yang 0.675993
…… …… ……
TABLE 3
Due to lawyers' excellence in different types of casesThe length is different, the winning rate is different, therefore, the winning rate of the same kind of cases of the lawyer transacting users is calculated by the similarity described by the case and the case, and the evaluation item F of the lawyer winning rate is established3:
Figure GDA0003145365440000111
In the formula, x0jTotal number of cases, x, of the victory complaints belonging to case j handled by lawyersjTotal number of cases belonging to case j, s, handled by lawyersjAnd j is the similarity of case description and case description of the user.
In this embodiment, an evaluation item F of a lawyer "lawyer winning rate3Then, the lawyer F can be obtained by the above calculation3Is 0.907723.
In order to make up the incompleteness of evaluation of lawyer professional abilities from the three evaluation items, the patent introduces an expert scoring system to establish an evaluation item F of' expert scoring4. A plurality of lawyers are extracted at intervals to score experts, and the experts can score the lawyers from the aspects of credit conditions of the lawyers, performances in specific cases and the like to finally obtain a comprehensive score.
This embodiment sets the evaluation item F of "expert score" of all lawyers4The value is 0.9, and other values may be used depending on the actual situation.
Then, the 4 indexes of "lawyer working years", "case handling duration", "lawyer winning rate" and "expert score" are integrated to calculate the professional ability score of the lawyer.
Fi＝[F1 F2 F3F4][λ1 λ2 λ3 λ4]T
In the formula, FiIs the professional competence score of lawyer i, [ F1 F2 F3 F4]Is a score vector, [ lambda ]1 λ2 λ3 λ4]TIs the transpose of the weight vector.
In this case, weights [0.30,0.25,0.20,0.25] are given, and in the actual case, the weights can be adjusted according to the situation. Table 4 shows bar professional ability scores (ellipses are not shown, descending order).
Figure GDA0003145365440000112
Figure GDA0003145365440000121
TABLE 4
7. Obtaining lawyer service convenience score
Metrics measuring how convenient a lawyer services include the number of cases the lawyer is handling, how close the lawyer is to the user's geographic location. How many cases are being processed can reflect how busy the attorneys are; if the number is small, the lawyer is more energetic in processing the case of the user; if the number is large, the attorneys are distracted when dealing with cases, and the working efficiency is low. Thus, the evaluation item B of "the number of cases being accepted" is created1:
Figure GDA0003145365440000122
Where p is the number of cases being handled by the attorney.
Considering the convenience of backward cooperation between the user and the lawyer and the distance between the lawyer and the geographic position of the user is also a factor to be considered, and establishing an evaluation item B of' the distance between the lawyer and the user2:
Figure GDA0003145365440000123
In this case, since the database is from the bureau of a city where lawyers are located, assuming that the user is in the same city as all lawyers, all lawyers will have their evaluation item B2All are 1.
Then, the 2 indexes are integrated to calculate the service convenience score of the lawyer.
Bi＝[B1 B2][μ1 μ2]T
In the formula, BiScore for convenience of service for lawyer i, [ B1 B2]To score the vector, [ mu ]1 μ2]TIs the transpose of the weight vector.
In this case, a weight [0.7,0.3] is given, and in the actual case, the weight can be adjusted according to the situation.
8. User selection of lawyer
The user can input the weighted values of three indexes of lawyer professional matching degree, lawyer professional ability and service convenience according to the self requirement. The final fraction is:
Figure GDA0003145365440000131
in the formula, ScoreiIs the composite score of lawyer i, [ Zi Fi Bi]In order to be a score vector, the score vector,
Figure GDA0003145365440000132
is the transpose of the weight vector.
And (4) sorting according to the comprehensive scores, displaying the lawyers 20 th before the comprehensive scores on the interface, and selecting lawyers of the heart instrument by clicking the lawyers to view the representation of the lawyers. The attorney representation includes a static representation of the attorney and a dynamic representation of the attorney. The lawyer static figure comprises the name, sex, license number, contact way of the law and the like of the lawyer; the dynamic image includes the period of time the lawyer is working, the number of cases the lawyer is working on, the professional area of the lawyer, etc.
In this case, assuming the weighting values are [0.4,0.3,0.3], the attorneys who scored 10 times in total are shown in table 5.
Figure GDA0003145365440000133
TABLE 5
Referring to fig. 2, a lawyer precision recommendation system, the system comprising:
a database module: the system is used for storing lawyer basic information, case bases and an extended vocabulary set thereof;
the basic data processing module: comprises a lawyer information processing unit and a case and vocabulary expansion unit;
the lawyer information processing unit is used for correlating basic information of lawyers with basic information of cases, and calculating the working years of the lawyers, the working duration of the cases and the number of the cases being worked by the lawyers;
the case routing vocabulary extension unit is used for updating and acquiring case routing extension vocabularies;
the man-machine interaction module: the system comprises a user input unit and a lawyer recommendation display unit;
the user input unit is used for providing a lawyer recommendation index which is input by a user for case description, positioning and selecting, wherein the lawyer recommendation index can be selected from lawyer professional matching degree or lawyer comprehensive index; the lawyer comprehensive indexes comprise lawyer professional matching degree, lawyer professional ability and lawyer service convenience, and a user can input the weight values of the three indexes according to the requirement;
the lawyer recommendation display unit is used for displaying lawyers with the scores close to the front, and a user can check the representation of the lawyer by clicking the lawyer and select the lawyer of the heart instrument; the lawyer portrait comprises a lawyer static portrait and a lawyer dynamic portrait, wherein the lawyer static portrait comprises a lawyer name, a gender, a practice license number, a law contact information and the like; the bar chart comprises bar chart parameters, bar chart parameters and the like.
Lawyer recommendation module: and the system is used for calculating lawyer professional matching degree scores or lawyer comprehensive scores according to user selection and sorting based on the case situation description. The lawyer professional matching degree score and the lawyer comprehensive score are calculated according to the method.

Claims (8)

1. A lawyer accurate recommendation method, comprising the steps of:
acquiring case description input by a user, and performing word segmentation processing to acquire key words;
calculating the similarity between the case and the expansion vocabulary thereof and the key vocabulary, and acquiring the similarity between the case and the description of the case;
calculating professional matching degree scores of each lawyer under the case description input by the user based on the similarity between the case description and the case description, the total number of the cases transacted by each lawyer and the number of the cases transacted by each lawyer under each case description;
according to the score sorting, displaying lawyers with the scores close to the front on an interface to recommend to the user;
the method further comprises the following steps: integrating lawyer professional matching degree scores, lawyer professional ability scores and lawyer service convenience scores, calculating to obtain lawyer comprehensive scores, sorting according to the comprehensive scores, and displaying the lawyers with the scores higher in the front on an interface to recommend the lawyers to the user; wherein the lawyer professional ability score evaluation indexes comprise lawyer practice period, case handling duration, lawyer victory rate and expert score; the evaluation indexes of the lawyer service convenience score comprise the number of cases being accepted by the lawyer and the distance between the lawyer and the geographic position of the user; the calculation method of the lawyer professional ability score is as follows:
using logarithmic function to quantify the relationship between the law enforcement period and the score, and establishing an evaluation item F of the law enforcement period1:
F1＝ln(tz+2)
In the formula, tzLaw enforcement years, tz+2 to avoid the case of attorneys with a practice age of one year and less than one year with a score of zero for the evaluation item;
calculating the handling time of lawyers for handling similar cases of the user, and establishing an evaluation item F of the' case handling time2：
Figure FDA0003145365430000011
Wherein m is the number, xjThe law is dealt with by a schoolerTotal number of cases, t, of jskThe handling duration of the kth case in the case group j, sjThe similarity of case description information of case j and the user is set;
calculating the winning rate of the lawyer handling the same type case of the user and establishing an evaluation item F of the lawyer winning rate3：
Figure FDA0003145365430000012
In the formula, x0jTotal number of cases submitted by lawyers belonging to the victory of case j;
introducing an expert rating system to establish an evaluation item F of' expert rating4Extracting a plurality of lawyers at intervals to score experts, and scoring the lawyers by the experts according to the credit conditions of the lawyers and the performances in specific cases to finally obtain a comprehensive score;
integrating the 4 indexes to calculate the professional ability score of the lawyer;
Fi＝[F1 F2 F3 F4][λ1 λ2 λ3 λ4]T
in the formula, FiIs the professional competence score of lawyer i, [ F1 F2 F3 F4]Is a score vector, [ lambda ]1 λ2 λ3 λ4]TIs the transpose of the weight vector.
2. A lawyer accurate recommendation method as claimed in claim 1, wherein said word segmentation process obtains key words as follows:
adding a judicial vocabulary into a custom dictionary of a word segmentation tool, and segmenting the text described by the case to obtain segmented words; further, the nouns, verbs and vernouns are reserved, repeated vocabularies are deleted, and key vocabularies are obtained.
3. A lawyer accurate recommendation method as claimed in claim 1 or 2, wherein the similarity between the acquired case and the case description is as follows:
let W1For a case in a database and its extended vocabulary set, W2Describing a set of key words for a user case, W1＝{w11,w12,…,w1eWherein e is W1Number of words contained, w11,w12,w1eAre respectively W11, 2, e of (1), W2＝{w21,w22,…,w2fWherein f is W2Number of key words involved, w21,w22,w2fAre respectively W21, 2 and f vocabularies in the specification;
W1and W2Similarity is as
Figure FDA0003145365430000021
Wherein Sim (w)1h，w2g) Obtained by a semantic model, h ═ 1, 2.., e; g 1, 2, ·, f;
then, the similarity Sim (W) is calculated1，W2) And (6) carrying out normalization processing.
4. A lawyer accurate recommendation method as claimed in claim 3, wherein the case and its extended vocabulary set are obtained by the following method:
a large number of various types of judicial texts and social news are obtained as linguistic data, a semantic model is trained by using a language model, case-based entries are expanded by combining the judicial entries and the semantic model, and the case-based entries and the extended vocabulary sets are correspondingly stored in a database to form a case and an extended vocabulary set thereof.
5. A lawyer accurate recommendation method as claimed in claim 1 or 2, wherein the professional matching degree score of each lawyer under the case description inputted by the user is calculated as follows:
assuming a total of n lawyers and m cases, the calculation expression is as follows:
Figure FDA0003145365430000022
wherein xijNumber of cases belonging to case j, s, representing lawyer i transactsjSimilarity of case description of user by j, NiRepresents the total number of cases, Z, of bar iiRepresents the professional matching degree score of lawyer i, wherein i is more than or equal to 1 and less than or equal to n and j is more than or equal to 1 and less than or equal to m.
6. A lawyer accurate recommendation method as claimed in claim 1, wherein the lawyer service convenience score is calculated as follows:
establishing an evaluation item B of' number of cases being accepted1:
Figure FDA0003145365430000023
Wherein p is the number of cases being processed by the lawyer;
establishing an evaluation item B of' lawyer and user far and near2:
Figure FDA0003145365430000024
Then, integrating the 2 indexes to calculate the service convenience score of the lawyer;
Bi＝[B1 B2][μ1 μ2]T
in the formula, BiScore for convenience of service for lawyer i, [ B1 B2]To score the vector, [ mu ]1 μ2]TIs the transpose of the weight vector.
7. The method as claimed in claim 1, wherein the calculation of the lawyer comprehensive score can input weighted values of three indexes of lawyer professional matching degree, lawyer professional ability and service convenience according to the user requirement, and the final score is:
Figure FDA0003145365430000025
in the formula, ScoreiIs the composite score of lawyer i, [ Zi Fi Bi]In order to be a score vector, the score vector,
Figure FDA0003145365430000026
is the transpose of the weight vector.
8. A recommendation system implemented by a lawyer accurate recommendation method according to claim 1, the system comprising:
a database module: the system is used for storing lawyer basic information, case bases and an extended vocabulary set thereof;
the basic data processing module: comprises a lawyer information processing unit and a case and vocabulary expansion unit;
the lawyer information processing unit is used for correlating basic information of lawyers with basic information of cases, and calculating the working years of the lawyers, the working duration of the cases and the number of the cases being worked by the lawyers;
the case routing vocabulary extension unit is used for updating and acquiring case routing extension vocabularies;
the man-machine interaction module: the system comprises a user input unit and a lawyer recommendation display unit;
the user input unit is used for providing a lawyer recommendation index which is input by a user for case description, positioning and selecting, wherein the lawyer recommendation index can be selected from lawyer professional matching degree or lawyer comprehensive index; the lawyer comprehensive indexes comprise lawyer professional matching degree, lawyer professional ability and lawyer service convenience, and a user can input the weight values of the three indexes according to the requirement;
the lawyer recommendation display unit is used for displaying lawyers with the scores close to the front, and a user can check the representation of the lawyer by clicking the lawyer and select the lawyer of the heart instrument; the lawyer portrait comprises a lawyer static portrait and a lawyer dynamic portrait, wherein the lawyer static portrait comprises a lawyer name, a gender, a practice license number and a connection mode of a law; the bar chart dynamic portrait comprises bar chart practice years, the number of cases being handled by bar charts and bar chart professional fields;
lawyer recommendation module: and the system is used for calculating lawyer professional matching degree scores or lawyer comprehensive scores according to user selection and sorting based on the case situation description.
CN201910573715.9A 2019-06-28 2019-06-28 Lawyer accurate recommendation method and system Active CN110347814B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910573715.9A CN110347814B (en) 2019-06-28 2019-06-28 Lawyer accurate recommendation method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910573715.9A CN110347814B (en) 2019-06-28 2019-06-28 Lawyer accurate recommendation method and system

Publications (2)

Publication Number Publication Date
CN110347814A CN110347814A (en) 2019-10-18
CN110347814B true CN110347814B (en) 2021-09-14

Family

ID=68176901

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910573715.9A Active CN110347814B (en) 2019-06-28 2019-06-28 Lawyer accurate recommendation method and system

Country Status (1)

Country Link
CN (1) CN110347814B (en)

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111159373B (en) * 2019-12-26 2023-04-07 上海智臻智能网络科技股份有限公司 Knowledge base establishing method and device for intelligent question-answering system and storage medium
CN111639256B (en) * 2020-04-20 2024-06-04 广东德诚科教有限公司 Discipline-based professional recommendation method, device, computer equipment and storage medium
CN111210172B (en) * 2020-04-21 2020-08-21 成都派沃特科技股份有限公司 Device for dispatching police officers
CN111861188A (en) * 2020-07-15 2020-10-30 中国工商银行股份有限公司 Case distribution method and device
CN113407729B (en) * 2021-05-11 2022-06-24 银江技术股份有限公司 Judicial-oriented personalized case recommendation method and system
CN113554532B (en) * 2021-06-16 2024-06-07 贝壳找房(北京)科技有限公司 Broker list page ordering method, storage medium and program product
CN113902290B (en) * 2021-09-14 2022-11-04 中国人民解放军军事科学院战略评估咨询中心 Expert matching effectiveness measuring and calculating method facing evaluation task
CN114003705B (en) * 2021-10-29 2024-09-20 平安国际智慧城市科技股份有限公司 Lawyer recommending method, device, equipment and storage medium based on data analysis
CN115759692A (en) * 2022-12-01 2023-03-07 徐玉峰 Moderator allocation method, system, medium and equipment based on two-dimensional code

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TW201530477A (en) * 2014-01-28 2015-08-01 Univ Southern Taiwan Sci & Tec Method, computer program and computer readable recording medium for recommending legal experts
CN105718526A (en) * 2016-01-15 2016-06-29 上海律巢网络科技有限公司 Data search method, device and system based on lawyer information
CN107563912A (en) * 2017-08-29 2018-01-09 广东蔚海数问大数据科技有限公司 A kind of lawyer recommends method and system
CN109409645A (en) * 2018-09-07 2019-03-01 平安科技（深圳）有限公司 The method and storage medium that electronic device, lawyer recommend

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TW201530477A (en) * 2014-01-28 2015-08-01 Univ Southern Taiwan Sci & Tec Method, computer program and computer readable recording medium for recommending legal experts
CN105718526A (en) * 2016-01-15 2016-06-29 上海律巢网络科技有限公司 Data search method, device and system based on lawyer information
CN107563912A (en) * 2017-08-29 2018-01-09 广东蔚海数问大数据科技有限公司 A kind of lawyer recommends method and system
CN109409645A (en) * 2018-09-07 2019-03-01 平安科技（深圳）有限公司 The method and storage medium that electronic device, lawyer recommend

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
"基于规则的句子相似度计算";张子宪等;《中国科技论文》;20170731;第1651-1674页 *

Also Published As

Publication number Publication date
CN110347814A (en) 2019-10-18

Similar Documents

Publication Publication Date Title
CN110347814B (en) Lawyer accurate recommendation method and system
Devi et al. Decision Support Systems Assessment of the best village in Perbaungan sub-district with the Simple Additive Weighting (SAW) Method: Decision Support Systems Assessment of the best village in Perbaungan sub-district with the Simple Additive Weighting (SAW) Method
CN108628833B (en) Method and device for determining summary of original content and method and device for recommending original content
CN107851097B (en) Data analysis system, data analysis method, data analysis program, and storage medium
CN109543178B (en) Method and system for constructing judicial text label system
CN111401040B (en) Keyword extraction method suitable for word text
Satria et al. Application of SAW in the Class Leader Selection Decision Support System
CN105653671A (en) Similar information recommendation method and system
CN110120001B (en) Method and system for scoring based on combination of knowledge graph library and memory curve
CN102663129A (en) Medical field deep question and answer method and medical retrieval system
KR101426765B1 (en) System and method for supplying collaboration partner search service
KR20080021017A (en) Comparing text based documents
CN106294744A (en) Interest recognition methods and system
CN109522397B (en) Information processing method and device
Ng et al. CrsRecs: A personalized course recommendation system for college students
CN114818678A (en) Scientific and technological achievement management method and platform and electronic equipment
CN112184021A (en) Answer quality evaluation method based on similar support set
CN109977214A (en) A kind of online knotty problem answer recommendation interactive approach and system of education of middle and primary schools
CN110781300A (en) Tourism resource culture characteristic scoring algorithm based on Baidu encyclopedia knowledge graph
CN110347812A (en) A kind of search ordering method and system towards judicial style
CN112395484A (en) User satisfaction evaluation method for automatically driving automobile
Nisa et al. Hybrid Method for User Review Sentiment Categorization in ChatGPT Application Using N-Gram and Word2Vec Features
CN112765311A (en) Method for searching referee document
CN114385927A (en) Scientific research collaborator recommendation method based on multi-similarity fusion
Ali et al. Manhattan Distance and Dice Similarity Evaluation on Indonesian Essay Examination System

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
CP01 Change in the name or title of a patent holder
CP01 Change in the name or title of a patent holder

Address after: 310012 1st floor, building 1, 223 Yile Road, Hangzhou City, Zhejiang Province

Patentee after: Yinjiang Technology Co.,Ltd.

Address before: 310012 1st floor, building 1, 223 Yile Road, Hangzhou City, Zhejiang Province

Patentee before: ENJOYOR Co.,Ltd.