WO2022262752A1 - 基于数据交互的信息推荐方法、装置、设备及存储介质 - Google Patents

基于数据交互的信息推荐方法、装置、设备及存储介质 Download PDF

Info

Publication number
WO2022262752A1
WO2022262752A1 PCT/CN2022/098860 CN2022098860W WO2022262752A1 WO 2022262752 A1 WO2022262752 A1 WO 2022262752A1 CN 2022098860 W CN2022098860 W CN 2022098860W WO 2022262752 A1 WO2022262752 A1 WO 2022262752A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
agent
historical
matching
agency
Prior art date
Application number
PCT/CN2022/098860
Other languages
English (en)
French (fr)
Inventor
蒙天庆
Original Assignee
诺正集团股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 诺正集团股份有限公司 filed Critical 诺正集团股份有限公司
Priority to EP22824229.3A priority Critical patent/EP4358005A1/en
Publication of WO2022262752A1 publication Critical patent/WO2022262752A1/zh

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
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/25—Integrating or interfacing systems involving database management systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/906—Clustering; Classification
    • 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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • 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
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0631—Item recommendations
    • 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
    • G06Q50/184—Intellectual property management

Definitions

  • the present application relates to the technical field of information processing, in particular to an information recommendation method, device, device and storage medium based on data interaction.
  • patents have become an important competitive resource for enterprises.
  • a patent document is a professional legal document. Applicants often need to entrust a patent agency to apply for a patent. The writing of the patent document is done by the patent attorney in the agency.
  • This application aims to solve the above-mentioned problem of inaccurate recommendation, and proposes an information recommendation method, device, equipment, and storage medium based on data interaction, which can more accurately match patent attorneys.
  • the first aspect proposes an information recommendation method based on data interaction, which is applied to the information processing platform, including:
  • the basic information of the customer includes: applicant information;
  • the agent list including a plurality of matching agent information
  • the second aspect proposes an information recommendation device based on data interaction, including:
  • An acquisition module configured to acquire the basic information of the client, where the basic information of the client includes: applicant information;
  • connection module configured to establish a connection with a third-party database, where patent case information is stored in the third-party database
  • An analysis module configured to use a preset patent analysis model to analyze all patent case data corresponding to the applicant information, and analyze and obtain historical application information corresponding to the applicant information;
  • the obtaining module is also used to obtain the client's agency demand information
  • a matching module configured to match the agent demand information and the historical application information with the agent information in the database to generate a matching agent list, the agent list including a plurality of matching agent information;
  • a returning module configured to return the generated list of agents to the client terminal.
  • the third aspect provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the processor performs the following steps:
  • the basic information of the customer includes: applicant information;
  • the agent list including a plurality of matching agent information
  • the fourth aspect provides a computer device, including a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor Perform the following steps:
  • the basic information of the customer includes: applicant information;
  • the agent list including a plurality of matching agent information
  • the above-mentioned information recommendation method, device, equipment and storage medium based on data interaction first, obtain customer information (applicant information), establish a connection with a third-party database, and then analyze and extract all patent case data corresponding to the applicant information to extract the history After applying for information, it is necessary to obtain the client's agency demand information, and then perform agent information matching based on historical application information and agency demand information, generate a matching agent list, and return the generated agent list to the client terminal.
  • the information recommendation method based on data interaction realizes the accurate recommendation of agents for customers through multi-faceted information matching.
  • the recommendation method is not limited to the recommendation within a certain agency, but is applicable to the recommendation for all agents, and has a wide range of applications.
  • Fig. 1 is a flowchart of an information processing method based on data interaction in an embodiment
  • Fig. 2 is a flow chart of a method for determining the professional level of an agent in one embodiment
  • FIG. 3 is a flowchart of an information recommendation method based on data interaction in an embodiment
  • Figure 4 is a flow chart of a method for calculating the degree of matching with each candidate agent in one embodiment
  • Fig. 5 is a flowchart of an information interaction method in an embodiment
  • FIG. 6 is a sequence diagram of an information interaction process in an embodiment
  • Fig. 7 is a structural block diagram of an information processing device based on data interaction in an embodiment
  • Fig. 8 is a structural block diagram of an information recommendation device based on data interaction in an embodiment
  • Fig. 9 is a structural block diagram of an information interaction device in an embodiment
  • Figure 10 is a block diagram of the internal structure of a computer device in one embodiment.
  • Step 102 acquiring the basic information of the agent, the basic information includes: identification information of the agent.
  • the agent here refers to the person engaged in patent agency work, also known as "patent attorney” or "patent attorney”.
  • the agent’s basic information includes the agent’s identification information, which is suitable for determining the agent’s identity.
  • the agent’s name can be directly used as the agent’s identity information, or the agent’s name can be used.
  • the person's agent qualification certificate number is used as the agent's identification information.
  • the agent's name + practice experience is used as the agent's identification information, wherein the practice experience includes the name of the agency corresponding to the agent during each practice period.
  • the agent’s basic information may also include other personal information, such as gender information, education information, professional information, agent qualification certificate information, agent practice certificate information, and agent practice certificate information.
  • the practice experience includes: when and which agency the agent practiced, for example, the agent Li San practiced in agency A during the period of 2012.1-2018.7, and practiced in agency B during the period of 2018.8-2021.4.
  • the professional direction of the agent At present, the industry is mainly divided into four professional agency directions: iSoftStone, electronics, machinery and biochemistry.
  • the agent’s basic information can be obtained directly from the agent’s registration information.
  • the agent needs to register with a real name on the information processing platform and fill in the Personal Information.
  • the agent's basic information can be captured by interacting with a third-party platform, for example, the agent's practice experience can be found from the "All China Attorneys Association website", from “ “Xuexin Net” can find the agent's education information, professional information, etc.
  • Step 104 establishing a connection with a third-party database, where patent case information is stored, and the patent case information includes: agent identification information.
  • the third-party database refers to a database storing patent case information.
  • the third-party database can be the patent retrieval database of the State Intellectual Property Office, of course, it can also be other patent retrieval databases, such as Wisdom Bud, Baiteng, incopat, etc.
  • the patent case information includes the agent's identification information, for example, each patent case information includes the name of the agency and the name of the agent.
  • Step 106 adopt the established patent case data processing model to process all the patent case data corresponding to the agent, and calculate the historical authorization rate of the agent within a preset time.
  • the patent case data processing model is used to batch process all the patent case data corresponding to the agent.
  • the legal status of each patent case can be extracted through the patent case data processing model.
  • the legal status includes: authorization, rejection, trial , Withdraw the four states.
  • withdrawal is divided into voluntary withdrawal and deemed withdrawal.
  • the preset time can be customized. For example, the time since the agent has been in business can be used as the preset time, or the last five years can be used as the preset time.
  • the calculation of the grant rate is only based on statistical calculations of invention patent cases. Since utility model and appearance patents do not involve substantive examination, the grant rate of utility model and appearance patents cannot accurately reflect the professional level of the attorney. Specifically, the patent case data processing model can first extract invention patents by case type, and then extract the legal status of each invention patent, and calculate the historical authorization rate of the agent based on the legal status of each invention patent. Statistics on the authorization rate based on invention patents are conducive to accurately evaluating the professional level of agents.
  • a preset number for example, 100
  • the number of rejections and the number of withdrawals are calculated to obtain the historical authorization rate that best reflects the current professional level of the attorney.
  • Step 108 extracting a preset number of patent cases to be evaluated within a preset period of time from all the patent case data corresponding to the agent.
  • the patent cases to be evaluated are randomly selected cases for experts to evaluate the quality of the attorney's writing. Since the professional level of an agent generally improves gradually with the accumulation of experience, in order to truly reflect the current professional level of an agent, the data of patent cases within a preset period (for example, the past three years) can be used A preset number (for example, 3) of patent cases are extracted as patent cases to be evaluated, which is conducive to an accurate evaluation of the current professional level of the patent attorney.
  • Step 110 send the patent case to be evaluated to the corresponding expert for review, and receive the returned review result.
  • the division of technical fields can be divided according to specific technologies. Specifically, the determination of technical fields can be determined according to technical classification numbers, and different technical classification numbers themselves represent the technical fields to which they belong. For example, G01 represents the field of measurement and testing. G02 means optics.
  • the determination of the technical field can be determined according to the technical classification number and the technical keyword at the same time, and the combination of the technical classification number + the technical keyword can obtain a more subdivided technical field. For example, if the technical classification number indicates program control, and the technical keyword is temperature control, then the determined technical field is: temperature control using software.
  • Step 112 Determine the professional level of the agent according to the agent's historical authorization rate and review results according to a preset algorithm.
  • the agent's historical authorization rate can accurately reflect the professional level
  • the review results can accurately reflect the quality of case writing
  • the combination of the two results can accurately realize the professional classification of the agent.
  • the two can be integrated by means of weighted summation.
  • the years of service of the agent and the total number of historical case applications are obtained, and the number of years of service and the total number of historical case applications are used as the input of the weight determination model.
  • the weight determination model is trained using a deep neural network model learned. Obtain the first weight of the historical authorization rate and the second weight of the review result corresponding to the agent output by the weight determination model; determine the professional level of the agent according to the historical authorization rate and the first weight, the review result and the second weight.
  • historical service evaluation information corresponding to the agent may also be acquired.
  • the historical service evaluation information refers to the evaluation information of the agent’s professionalism and service by the customers of the agent’s historical service.
  • the historical service evaluation information can be a score or a grade. For example, the grade can be divided into very satisfied, satisfied , Average, Dissatisfied.
  • the agent's historical service evaluation information can truly reflect the agent's service quality, including: service attitude and professional recognition. Then, according to the agent's historical authorization rate, review results and historical evaluation information, the agent's professional level is determined according to a preset algorithm. That is, the results of the three are combined to classify, and this method can more accurately realize the professional classification of agents.
  • the obtained historical service evaluation information may be empty.
  • the historical evaluation information can be set as a default value, or the weight of this item can be set to 0, and the weights of the other two items can be increased.
  • the classification of professional levels can be customized according to needs. For example, professional levels can be divided into five levels, namely AAA, AA+, AA, AA-, and A. Of course, 6 levels can also be set.
  • the expression form of the level can also be customized. For example, it can also be displayed by lighting up a few stars.
  • Step 114 generating personal introduction information corresponding to the agent according to the professional level of the agent according to a preset template.
  • the personal introduction information of the agent includes the professional level of the agent, so that the professional level of the agent can be reflected more intuitively.
  • the generated personal introduction information of the patent attorney not only includes the professional level that can reflect the professional level of the attorney, but also includes: the main applicants represented by the attorney in the past and the main technical fields.
  • other information may also be included, for example, including the professional agency direction of the agent.
  • the industry is mainly divided into four professional agency directions of iSoftStone, electronics, machinery and biochemistry. Of course, it can also be adjusted according to changes in the actual situation, for example, to further refine the corresponding agency directions.
  • the main applicants represented by the agent and the main technical fields represented by the agent are obtained in the following way: analyze and process all the patent case data corresponding to the agent, and obtain the applicants represented by the agent historically and the technical field of historical proxies.
  • the main applicant refers to sorting the number of cases of each applicant represented in the past from the most to the least, and selecting the applicants with the preset number as the main applicant. For example, if you represent Company A with 10 patents, Company B with 20 patents, Company C with 34 patents, Company D with 5 patents, and Company E with 12 patents, then the applicants are sorted according to the number of patent cases, namely Company C, Company E, and Company E. Company B, Company E, Company A and Company D, and then select the first three applicants as the main applicants, namely Company C, Company B and Company E.
  • the main technical field refers to sorting the number of cases in various technical fields represented by history, and selecting the technical field with the preset number (for example, the first three) as the main technical field. By selecting the main applicant and the main technical field, it is convenient to display the agent information in a more focused manner.
  • the above method includes: sending the generated personal introduction information to a terminal for display. That is, after the terminal logs in to the information processing platform, it can view the personal introduction information of the agent. Solved the problem that the personal introduction information of the agent cannot be obtained accurately and quickly in the industry.
  • the above-mentioned information processing method, device, equipment and storage medium based on data interaction first obtain all the patent case data represented by the agent in history through interaction with the third-party database according to the basic information of the agent, and then analyze all the patent cases corresponding to the agent Analyze the case data and calculate the historical authorization rate of the agent. Further, randomly select a preset number of patent cases to be evaluated within the preset number of years, send the patent cases to be evaluated to experts in the corresponding technical field for review, and receive the returned review As a result, the agent's professional level is finally calculated based on the agent's corresponding historical authorization rate and review results.
  • the real patent case data of the agent's historical representation can be captured, based on the real patent case data, the real historical authorization rate of the agent can be obtained, and by extracting the real patent case Expert review, to get real review results.
  • the professional level of the agent is determined through the comprehensive historical authorization rate and review results corresponding to the agent, and the determination of the professional level is authentic and credible.
  • the personal introduction information of the agent is generated according to the preset template. The generated personal introduction information of the agent is not only authentic and reliable, but also comprehensive, so that the applicant can comprehensively And accurately understand the professional level of the agent.
  • the personal introduction information of the attorney is open to the outside world, and the applicant can quickly obtain it on the terminal, and what the applicant can view is the personal introduction information of all patent attorneys, not limited to only one patent attorney. Agency patent attorney.
  • the identification information of the agent includes: the name of the agent and the execution experience, the practice experience includes the name of the agency corresponding to the agent during the practice period; the patent case data processing model established in the adoption Before processing all the patent case data corresponding to the agent, the method further includes: extracting all the patent case data corresponding to the agent from the third-party database according to the name of the agent and the practice experience data.
  • the basic information includes: professional agency direction; extracting a preset number of patent cases to be evaluated within a preset number of years from all patent case data corresponding to the agent, including: all patent cases corresponding to the agent Candidate patent case data that matches the direction of professional agency is screened out from the data; patent cases to be evaluated are extracted from the candidate patent case data.
  • the professional agency direction refers to the agency direction that the agent is good at, which is mainly divided into four professional agency directions: iSoftStone (software communication), electronics, machinery and biochemistry (biology and chemistry).
  • iSoftStone software communication
  • electronics electronics
  • machinery biochemistry
  • biochemistry biochemistry and chemistry
  • the established patent case data processing model is used to process all the patent case data corresponding to the agent, and the historical authorization rate of the agent is calculated, including: using the established patent case data processing model to extract all the patent case data Legal status, the legal status includes: one of authorization, trial, rejection and withdrawal, the patent case data processing model is used to determine the regular expression corresponding to the legal status, and extract the legal status of each case based on the regular expression; Count the number of authorized cases whose legal status is authorized, the number of rejected cases whose legal status is rejected, and the number of withdrawn cases whose legal status is withdrawn; calculate the historical authorization rate based on the number of authorized cases, the number of rejected cases, and the number of withdrawn cases.
  • counting the number of authorized cases whose legal status is authorized, the number of rejected cases whose legal status is in dismissed state, and the number of withdrawn cases whose legal status is in withdrawn state includes: counting the preset number of cases closest to the current time The number of authorized cases, the number of rejected cases and the number of withdrawn cases in the closed cases; the historical authorization rate is calculated according to the number of authorized cases, the number of rejected cases and the number of withdrawn cases, including: the number of closed cases based on the preset number closest to the current time The number of authorized cases, the number of rejected cases and the number of withdrawn cases are calculated to obtain the latest historical authorization rate.
  • closed cases include authorized cases, dismissed cases and withdrawn cases.
  • the authorization rate of the preset number for example, 100
  • the historical authorization rate is dynamically calculated, and as time goes by, the historical authorization rate is constantly updated, so that the professional level of the agent can be reflected in time.
  • counting the number of authorized cases whose legal status is authorized, the number of rejected cases whose legal status is rejected, and the number of withdrawn cases whose legal status is withdrawn includes: Candidate patent cases; count the number of authorized cases in the authorized state, the number of rejected cases in the rejected state and the number of withdrawn cases in the withdrawn state among the candidate patent cases; calculate the historical authorization rate based on the number of authorized cases, the number of rejected cases and the number of withdrawn cases , including: calculating the historical authorization rate corresponding to the preset time period based on the number of authorized cases in the authorized state, the number of rejected cases in the rejected state, and the number of withdrawn cases in the withdrawn state among the candidate patent cases.
  • the authorization rate of the cases within the preset time period is counted.
  • the preset time period can refer to the time period when the application date falls within a specific year. Since the patent examination cycle is relatively long, generally taking more than one year or even 3 years, it is possible to screen out the cases with the application date in the past three years, and then Statistics on the authorization of candidate patent cases in the past three years, including: the number of authorized cases, the number of rejected cases and the number of withdrawn cases, which can reflect the professional level of the agent in the past three years, and the calculated authorization rate closer to the agent's current professional level.
  • the agent's professional level is determined according to a preset algorithm according to the agent's historical authorization rate, review results, and historical service evaluation information, including:
  • Step 202 obtain the attorney's years of practice and the total number of historical case applications, and use the years of practice and the total number of historical case applications as inputs to the weight determination model.
  • the weight determination model is obtained by training and learning with a deep neural network model.
  • the weight determination module is obtained by training and learning based on the deep neural network model.
  • the training weight determination model adopts a supervised training method, and training data needs to be constructed. Each weight manually marked is used as a label for training, and then based on the set loss function, the gradient descent method is used to continuously adjust the weights to determine the parameters in the model until the model reaches the convergence condition.
  • Step 204 obtaining the first weight of the historical authorization rate corresponding to the agent output by the weight determination model, the second weight of the review result and the third weight of the historical service evaluation information.
  • the first weight, the second weight and the third weight of the model output are determined by the weight obtained.
  • the first weight reflects the influence degree of the historical authorization rate on the professional level of the agent
  • the second weight reflects the influence of the review result on the professional level of the agent.
  • the degree of influence, the third weight reflects the degree of influence of historical evaluation information on the agent's professional grade.
  • the weight of the review results is relatively greater for the agent with shorter working experience, and the weight of historical authorization rate and historical service evaluation information is smaller.
  • the rate is often not high, or there is no relevant data.
  • the corresponding historical service evaluation information will be less, so the reference value is not very meaningful.
  • the weight of historical authorization rate and historical service evaluation information will increase relatively.
  • Step 206 determine the professional level of the agent according to the historical authorization rate and the first weight, the review result and the second weight, historical service evaluation information and the third weight.
  • the professional level of the agent can be determined by weighted summation.
  • the historical authorization rate, review results, and historical service evaluation information can all be converted into scores, and after the comprehensive score is obtained, the professional level of the agent is determined based on the comprehensive score. For example, there are five levels.
  • all patent case data corresponding to the attorney is extracted from the third-party database according to the attorney's name and practice experience, including: determining the target search condition according to the attorney's name and practice experience, and the target search condition includes the attorney The name of the person, the name of the practicing agency and the corresponding practice time; all the patent case data corresponding to the agent are extracted from the third-party database according to the target search criteria.
  • the target search conditions must include the name of the agent + the name of the practicing agency + the corresponding practice time.
  • the target search condition By setting this target search condition, the interference caused by the same name and the same surname can be eliminated as much as possible. Therefore, it is more conducive to accurately obtain all patent case data corresponding to the agent.
  • the technical analysis of the patent case to be evaluated is carried out to determine the technical field to which the patent case to be evaluated belongs, including: obtaining the technical classification number and field keyword corresponding to the patent case to be evaluated; determining according to the technical classification number and field keyword The technical field to which the patent to be evaluated belongs.
  • the field keyword refers to the field where the technology is applied, and the technical classification number is used to determine the technical classification.
  • the specific technical field can be determined according to the technical classification number and the field keyword.
  • the temperature control is determined according to the technical classification number.
  • field keywords are: electronic cigarettes, then it is determined that the patent to be evaluated belongs to the temperature control of electronic cigarettes.
  • the data of all patent cases corresponding to the agent are analyzed and processed, and the applicants represented by the agent and the technical fields represented by the history are obtained statistically, including: grabbing the classification number information and summary information of each patent case , analyze the abstract information, and extract the technical keywords; use the classification number information and technical keywords as the input of the technical field classification model, determine the technical field category corresponding to the patent case, and use the determined technical field category as the technical field of the patent case Label.
  • the classification number information can be specific to the group, for example, G06K9/00.
  • the abstract information refers to the abstract content in the patent document.
  • the abstract information is analyzed to extract technical keywords, and the extraction of technical keywords can be extracted based on semantics, that is, the technical keywords are extracted through semantic analysis of the abstract information.
  • use the technical field classification model to determine the technical field category of the corresponding patent case, and store the technical field category as the technical label of the patent case, so that it is not only convenient for the applicant to view the historical information of the technical field represented by the agent, but also convenient for follow-up Make matching recommendations in the field of patented technologies.
  • the technical field classification model includes: a first feature model, a second feature model and a classification model, the first feature model is used to determine the first feature vector corresponding to the classification number according to the classification number, and the second feature model is used to The second eigenvector corresponding to the technical keyword is determined, and the classification model is used to determine the category of the technical field corresponding to the patent case according to the first eigenvector and the second eigenvector.
  • the first feature model is used to convert the classification symbol into the first feature vector
  • the second feature model is used to convert the technical keywords into the second feature vector
  • the first feature vector and the second feature vector are combined, and the The combined feature vectors are used as the input of the classification model
  • the classification type is used to classify the technical field of the patent case according to the information contained in the first feature vector and the second feature vector.
  • the basic information also includes: the agent's qualification certificate number. After obtaining the agent's basic information, it also includes: obtaining the agent's official practice experience according to the agent's qualification certificate number; If the verification is successful, enter the step of establishing a connection with the third-party database; if the verification fails, return the result of verification failure.
  • the data of all patent cases corresponding to the agent are analyzed and processed, and the applicants represented by the agent and the technical fields represented by the history are obtained statistically, including: extracting the applicant information in each case, and counting the same The number of cases corresponding to the applicant information; determine the main applicant according to the number of cases corresponding to each applicant information, and the main applicant is the applicant with the highest number of cases; obtain the technical classification number and technical keywords of each case, according to The technical classification number and technical keywords determine the technical field corresponding to each case; the number of cases corresponding to the same technical field is counted, and the main technical field is determined based on the number of cases corresponding to the same technical field.
  • the main applicants represented by the agent before and the main technical fields represented by the agent are counted.
  • the main applicant refers to the main customer of the historical service
  • the main technical field refers to the main technical field of the historical service.
  • the agent mainly serves clients such as Company A, Company B, Company C, and Company D, etc.
  • the main technical fields are: Field I, Field II, Field III, etc.
  • subsequent customers can have a more accurate understanding of the agent based on the agent's personal introduction information, so as to choose the most suitable agent for them.
  • an information recommendation method based on data interaction is proposed, which is applied to the information processing platform, including:
  • step 302 the basic information of the customer is acquired, and the basic information of the customer includes: applicant information.
  • the applicant information may be a personal name or a business name.
  • the applicant information refers to the full name of the enterprise.
  • a customer account can be associated with one or more applicant information.
  • the customer's basic information may also include: enterprise qualification information, enterprise legal person information, and the like.
  • Step 304 establishing a connection with a third-party database, where patent case information is stored.
  • the third-party database refers to a database storing patent case information.
  • the third-party database can be the patent retrieval database of the State Intellectual Property Office, of course, it can also be other patent retrieval databases, such as Wisdom Bud, Baiteng, incopat, etc. For cases filed by agents, applicant information is included in the patent case information.
  • step 306 the preset patent analysis model is used to analyze all the patent case data corresponding to the applicant information, and the historical application information corresponding to the applicant information is obtained through analysis.
  • the patent analysis model is used to analyze all the captured patent case data to obtain the applicant's historical application status.
  • the historical application information includes: at least one of the technical field of the patent case of the historical application, the agency of the historical cooperation, and the agent of the historical cooperation.
  • the technical field of the patent case can be analyzed through the classification number, and then information such as the historical cooperation agency and agent can be extracted from the patent case data.
  • Obtaining the patent case data of the applicant's historical applications is conducive to analyzing and obtaining the actual needs of customers.
  • the technical field is determined by the classification code and the technical keywords extracted from the summary information.
  • Step 308 obtaining the customer's agency demand information.
  • the agency requirement information refers to the client's request information for the agent.
  • the agency requirement information includes but not limited to: one or more of field requirements, level requirements, professional requirements, agency requirements, time limit requirements, location requirements, and the like.
  • Field requirements refer to the requirements for the agent's professional agency direction, which can be divided into four directions: iSoftStone, electronics, machinery and biochemistry.
  • Level requirements refer to the level requirements for the agent. The higher the level of the agent, the better the writing quality of the agent, and of course the higher the corresponding agency fee.
  • Professional requirements refer to the requirements for the agent's major. For example, for customers in the LED field, they often require their agents to understand LED-related knowledge, so agents in the optoelectronic field are preferred.
  • the agency demand reflects the customer's trust in a certain agency. If the customer has cooperated horrin with a certain agency before, he can give priority to choosing an agent under the agency for cooperation.
  • the time limit requirement refers to the limitation on the time for the agent to return the manuscript. For example, for more urgent cases, the client requires the manuscript to be returned within a short time.
  • Location requirements refer to the requirements for the location of the agent. For example, some customers want to communicate with the agent face-to-face, and they will give priority to the agent who is closer to them.
  • Step 310 matching agent information in the database with agent demand information and historical application information to generate a matching agent list, which includes multiple matching agent information.
  • agent demand information and historical application information are integrated, and then matched with the agent information in the database to generate an agent list, which can be sorted according to the degree of matching from high to low, so as to facilitate subsequent customers to choose.
  • agent here refers to the person engaged in patent agency work, also known as "patent attorney” or "patent attorney”.
  • the agent list can also be sorted by multiple dimensions. For example, it can be divided into the professional dimension, that is, the highest professional matching degree, the level dimension, that is, the highest level matching degree, and the comprehensive dimension. That is to say, the dimensional ranking of multiple indicators is integrated, and customers can choose the sorting dimension according to the weight, which is conducive to providing customers with more flexible choices.
  • the historical evaluation information refers to the evaluation information of the client on the agent of the historical agency.
  • the historical evaluation information can reflect the customer's recognition of the agent of the historical agency. If the recognition is high, the agent can be recommended in the future. If the recognition is low, the agent will not be recommended in the future.
  • Step 312 returning the generated proxy list to the client terminal.
  • the agent list is returned to the client terminal, and the client makes a selection according to the agent list, which greatly improves the selection accuracy.
  • the above-mentioned information recommendation method, device, equipment and storage medium based on data interaction first, obtain customer information (applicant information), establish a connection with a third-party database, and then analyze and extract all patent case data corresponding to the applicant information to extract the history After applying for information, it is necessary to obtain the client's agency demand information, and then perform agent information matching based on historical application information and agency demand information, generate a matching agent list, and return the generated agent list to the client terminal.
  • the information recommendation method based on data interaction realizes the accurate recommendation of agents for customers through multi-faceted information matching.
  • the recommendation method is not limited to the recommendation within a certain agency, but is applicable to the recommendation for all agents, and has a wide range of applications.
  • the historical application information includes: the technical field of the patent cases of the historical application, the agency and the agent of the historical cooperation; all the patent cases corresponding to the applicant information are analyzed using the preset patent analysis model The data is analyzed, and before the analysis obtains the historical application information corresponding to the applicant information, it also includes: grabbing all the patent case data corresponding to the applicant information from the third-party database according to the applicant information.
  • the applicant information is used as the search condition, and all patent case data corresponding to the applicant are captured from the third-party database. In this way, the captured patent case data can be guaranteed to be accurate.
  • the preset patent analysis model is used to analyze all the patent case data corresponding to the applicant information, and the historical application information corresponding to the applicant information is analyzed, and the historical application information includes: the technical field of the patent case of the historical application , historical cooperation agencies and agents, including: using the preset patent analysis model to extract the technical classification number, technical keywords, agency name and agent corresponding to each patent case; according to the technical classification number and technical keywords Determine the technical field corresponding to the patent case; extract the agency corresponding to each patent case, and count the number of first cases handled by each agency in history, and sort the historical cooperation agencies according to the number of first cases; extract each patent For the agent corresponding to the case, the number of second cases represented by each agent is counted, and the historical cooperation agents are sorted according to the number of second cases.
  • the patent analysis model is used to extract the technical classification number, technical keywords, agency name and agent of each patent case.
  • the technical field is used to determine the field of the customer's technology, for example, whether it belongs to the field of electronic cigarettes or smart home. Then the agency information is used to obtain the agency that the customer prefers to cooperate with. Agent information is used to obtain the agent that the customer prefers to cooperate with.
  • the agency requirement information includes: at least one of field requirements, level requirements, professional requirements, and agency requirements
  • the agent information includes: level information, field information, professional information, and agency information
  • the agency demand information historical evaluation information and historical application information
  • match the agent information in the database to generate a matching agent list including: screening candidate agents that match the agency demand information based on the agency demand information and agent information,
  • the matching degree with each candidate agent is calculated, and the agent list is generated according to the matching degree.
  • the agent information in the agent list is arranged according to the matching degree.
  • the field can be divided according to the actual situation.
  • the field can be divided into four fields, namely, software communication, electronics, machinery and biochemistry.
  • the field can be further subdivided, for example, software communication can be further subdivided into: pure software, combination of software and hardware, communication, etc.
  • the level requirement refers to the professional level requirement of the agent, and the agent is divided into multiple professional levels in advance, and the professional level reflects the professional level of the agent.
  • Professional requirements refer to the majors the agent has studied, for example, majors are divided into optoelectronic communication, electromechanical automation, and so on.
  • Agency means the agency in which the agent works.
  • Agency requirements can be divided into two types, one is clear requirements, that is, the agency is designated, and the other is fuzzy requirements, which only need to specify the conditions that the agency must meet.
  • acquiring the customer's agency demand information includes: acquiring the agency demand content input by the customer, performing semantic analysis on the agency demand content, and extracting the semantic information of the agency demand content; determining the customer's agency demand information based on the semantic information,
  • the agency requirement information includes: at least one of domain requirement, professional level requirement, professional requirement and time limit requirement.
  • the content of the agent’s demand is the content that the customer needs to input.
  • the input method can be voice input or manual input.
  • the customer voice input “I need an agent in the electrical field, and request to return the manuscript within 15 days”.
  • the customer's agent demand information extracted is: electrical field, 15-day time limit.
  • performing semantic analysis on the content of the agency demand to extract the semantic information of the content of the agency demand includes: taking the content of the agency demand as the input of the semantic analysis model, and using the semantic analysis model to extract the semantic information of the content of the agency demand, Semantic information includes: semantic relationship and semantic content.
  • semantic information is often used to represent the intention information of the user's intention.
  • the intention information is displayed in the form of a triplet, a combination of triplets, a triplet of intentions, or a combination of triplets of intentions.
  • the semantic information includes triples or combinations of triples.
  • a triple refers to structural data in the form of (x, y, z), used to identify x, y, z and the corresponding relationship.
  • a triple consists of a syntactic/semantic relationship and two concepts, entities, words or phrases.
  • Intention triples are user intentions stored in the form of triples, which identify a small unit in a complete intention and can be identified as (subject, relation, object), where subject is the first entity, and relation represents subject and The relationship between objects, object represents the second entity.
  • subject is the first entity
  • relation represents subject
  • the relationship between objects, object represents the second entity.
  • I need electrical agents. Represented by triples (I, Requires Relation, Electron Agent).
  • the training of the semantic analysis model often requires the construction of a large amount of data. Since the application scenario of the semantic analysis model in this scheme is relatively special, the construction of the training data of the semantic analysis model has its particularity. Based on the particularity of the application scenario, the proposed A method for rapidly building training datasets for semantic analysis models. Determine candidate keywords for field requirements, candidate keywords for professional-level requirements, candidate keywords for professional requirements, and candidate keywords for time-limit requirements;
  • the training sentences containing the candidate keywords of the domain requirements are automatically generated, and then the corresponding candidate keywords of the domain requirements are used as the semantic annotation of the training sentences;
  • the training sentences containing the candidate keywords of the professional level requirements are automatically generated, and the corresponding candidate keywords are used as the semantic annotation of the training sentences;
  • the training sentences containing the candidate keywords of the professional requirements are automatically generated, and the corresponding candidate keywords are used as the semantic annotation of the training sentences;
  • a training sentence containing the candidate keyword of the time limit requirement is automatically generated according to a preset template, and the corresponding candidate keyword is used as a semantic annotation of the training sentence.
  • the agent information includes: professional level information, field information, professional information and time information; match the agent information in the database according to the agent demand information, historical evaluation information and historical application information to generate a matching agent List, including: screening candidate agents from agent information based on agent demand information; The matching degree generates an agent list, and the agent information in the agent list is arranged according to the degree of matching.
  • the information contained in the agency demand information is relatively accurate demand information, so the agent that matches the agency demand information can be selected as a candidate agent based on the agency demand information.
  • the agency demand information contains the agent
  • the agent who meets the professional level can be directly screened out as a candidate agent, which is beneficial to reduce the computational workload for subsequent matching.
  • the comprehensive matching degree is calculated according to the agent demand information, historical evaluation information and historical application information, and the matching degree with each candidate agent is obtained. Sorting according to the degree of matching is beneficial for customers to choose agents with high matching degree.
  • the candidate agents are first screened out based on the agent demand information, and then the matching degree with each candidate agent is calculated, which helps to reduce the workload of matching calculation and greatly improves the matching efficiency.
  • the matching degree with each candidate agent is calculated according to the agent demand information, historical evaluation information and historical application information, including:
  • Step 402 determine field requirements, level requirements, professional requirements and time limit requirements according to agency requirement information, historical evaluation information and historical application information.
  • the demand information contained in the agency demand information is relatively comprehensive, for example, it includes domain demand, agency demand, and professional demand, then such information can be determined according to the agency demand information. If the agency demand information does not contain so much information, it is necessary to analyze the missing information based on the historical application information and historical evaluation information. Assuming that there is no professional demand in the agency demand information, then the customer’s needs can be analyzed based on the technical fields in the historical application information. Professional needs information.
  • the level requirement can be empty, that is, the user can not limit the level.
  • the professional requirement can also be empty
  • the time limit requirement can also be empty, that is, the user can not limit the major and time limit.
  • the field requirement cannot be empty. The types of cases that are suitable for agents in different fields are different. Therefore, in order to match the client with a suitable field for the case, the field requirement cannot be empty.
  • Step 404 performing similarity calculation according to the domain requirements and the domain information corresponding to the candidate agent, to obtain the domain matching degree corresponding to the domain requirements.
  • the domain matching degree can be calculated by using the domain similarity degree.
  • the following formula is used to calculate the domain matching degree.
  • the domain requirements are expressed in the form of vectors.
  • the agent's domain information is also Expressed in vector form.
  • D represents the domain similarity
  • x i represents the i-th eigenvalue in the domain demand vector
  • y i represents the i-th eigenvalue in the domain information vector.
  • Step 406 Carry out correlation degree calculation according to the professional requirements and the professional information corresponding to the candidate agents, to obtain the professional matching degree corresponding to the professional requirements.
  • the professional matching degree can be calculated by professional similarity, and in one embodiment, can be calculated by the following formula:
  • d represents professional similarity
  • x i represents the i-th eigenvalue in the professional demand vector
  • y i represents the i-th eigenvalue in the professional information vector.
  • Step 408 matching according to the level requirement and the level information corresponding to the candidate agent, to obtain the level matching degree corresponding to the level requirement.
  • the matching rules between levels can be set in advance.
  • the matching degree of the same level is 100%, and the level of matching can be higher or lower.
  • the matching degree is set to 0, and the matching degree higher than the second level can be gradually reduced.
  • the matching degree between the third level and the second level is 80%, and the matching degree between the fourth level and the second level is 60%.
  • Step 410 matching the time limit requirement with the time information corresponding to the candidate agent to obtain the time limit matching degree corresponding to the time limit requirement.
  • the rules of the matching degree between the time limits can be set according to the needs, and the matching principle can be low or high. For example, if the customer requires a time limit of 10 days to return manuscripts, then the time information can be returned within less than 10 days. If it exceeds 10 days, the corresponding matching degree is set to 0.
  • Step 412 determine the matching degree with each candidate agent according to the field matching degree, professional matching degree, level matching degree and time limit matching degree.
  • determining the matching degree with each candidate agent according to the field matching degree, professional matching degree, level matching degree and time limit matching degree includes: obtaining customer preference information from agent demand information and historical evaluation information, Preference information includes: the client's preference for domain requirements, professional requirements, level requirements, and time limit requirements; use the preference information as the input of the weight analysis model to obtain the domain weight corresponding to the domain matching degree output by the weight analysis model, and the professional matching degree The corresponding professional weight, the level weight corresponding to the level requirement and the time limit weight corresponding to the time limit requirement; according to the field matching degree, field weight, professional matching degree, professional weight, level requirement, level weight, time limit requirement and time limit weight The degree of match for each candidate agent.
  • the agent information database and customer information database are pre-established on the information processing platform, so that when clients need to recommend an attorney, they can directly obtain the information required by the attorney from the Obtain historical application information from the customer information database,
  • an information interaction method applied to an information processing platform includes:
  • Step 502 establishing an agent information database, which includes the professional level of the agent.
  • the personal introduction information of the agent is stored in the agent information database, and the personal introduction information of the agent includes: professional level of the agent.
  • the agent's personal introduction information may also include: the main applicant of the historical agent and the main technical field of the historical agent. The agent's personal introduction information is obtained through the above-mentioned information processing method based on data interaction.
  • Step 504 establishing a customer information database, which includes: historical application information.
  • each customer information includes: historical application information
  • historical application information refers to the information obtained by analyzing the patent file data of the customer's historical application, including: patent cases of historical application At least one of the technology field, the agency with which the company has worked with, and the agent.
  • the customer information database further includes: historical evaluation information; the historical evaluation information refers to the customer's evaluation of the historical cooperation agent, including: evaluation of professionalism and evaluation of service attitude, etc.
  • the matching of the agent information in the agent information database according to the agent demand information, the historical application information corresponding to the target customer, and generating a matching agent list includes:
  • Step 506 Obtain the agent demand information of the target customer, match the agent information in the agent information database according to the agency demand information and the historical application information corresponding to the target customer, and generate a matching agent list, which contains multiple Matching agent information.
  • the agency demand information refers to the relatively accurate demand information input by the customer.
  • the agency demand information includes: requirements for the professional level of the agent, professional requirements for the agent, requirements for the practice years of the agent, and the time limit for the completion of the case. One or more of the requirements, etc. Synthesize agency demand information and historical application information, and then match it with the agent information in the database to generate an agent list, which can be sorted from high to low according to the degree of matching, so that subsequent customers can choose easily.
  • it also includes: obtaining the historical evaluation information of the target customer, matching the agent information in the agent information database according to the agent demand information and the historical application information and historical evaluation information corresponding to the target customer, and generating a matching Proxy list.
  • Step 508 returning the generated proxy list to the client terminal.
  • the agent list is returned to the client terminal, and the client makes a selection according to the agent list, which greatly improves the selection accuracy.
  • the agent information database and the customer information database are pre-established on the information processing platform.
  • the customer needs to recommend an agent, after obtaining the information required by the agent, he can directly obtain the historical application from the customer information database.
  • Information and historical evaluation information and then based on the agent demand information, historical application information and historical evaluation information to match the agent information in the agent information database to obtain the matching agent list, and then the agent list
  • the information is returned to the client terminal as a recommendation result.
  • the attorney information database and the client information database are pre-established, it is possible to quickly and accurately recommend matching patent attorneys for the client.
  • the above information interaction method further includes: receiving the intended agent selected from the agent list sent by the client terminal, and sending the client's intention request to the intended agent terminal; receiving the intention agent sent by the intended agent terminal Result; when the intention result is agreement, establish an instant messaging connection channel between the client terminal and the agent terminal, and the instant messaging connection channel is used for case communication between the client and the agent; the previous case communication process between the client and the agent Save it as a case communication process file, and the case communication process file is stored in association with the corresponding patent case.
  • the client selects an intended agent from the agent list. Since there is a two-way selection on the information processing platform, the intention request needs to be sent to the intended agent terminal for confirmation. Only after the intended agent agrees to entrust, the instant communication connection channel between the client terminal and the agent terminal will be established. , the client and the agent can communicate on the case through the instant messaging connection channel, and save the communication process file, which is convenient for review in case of subsequent disagreement, and it is also convenient for the agent to repeatedly review the communication process file to understand the plan. By associating and storing the case communication process files with the corresponding patents, it is also convenient for subsequent viewing of the communication records corresponding to the case, without the need to search through numerous information.
  • the information of all current instant messaging tools is stored uniformly. When a certain point in time or a certain event needs to be checked, it needs to be searched from many historical chat records, which is inefficient.
  • the case communication process files are innovatively stored in association with the corresponding patent cases, so that the communication process of each case can be quickly checked in the follow-up.
  • the above-mentioned information interaction method further includes: encrypting and saving the case communication process files to ensure the security of customer information.
  • the above method further includes: establishing a customer case information database, including: interacting with a third-party database, grabbing published patent case information corresponding to customer information from the third-party database, published patent Case information includes: case review status; obtain pending patent case information corresponding to customer information, and pending patent case information includes: case processing status; store published patent case information and pending case information corresponding to the same client To the customer case information database corresponding to the same customer, one customer corresponds to a customer case information database.
  • a customer case information database is established for each customer on the information processing platform, and the customer’s disclosed patent case data is captured from a third-party database and stored.
  • the pending patent case information corresponding to the customer is added to the customer case information database, so that the customer can quickly view the status of all cases by logging into the information processing platform.
  • customers can directly manage cases through the information processing platform, so that customers can understand the status of each case more quickly and accurately.
  • this method is undoubtedly faster and more accurate.
  • the above method further includes: receiving a case viewing request sent by the client terminal, obtaining the corresponding authorization information of the client in response to the request, obtaining the patent case information corresponding to the authorization information from the client case information database and returning it to the client terminal for further processing. exhibit.
  • the client terminal After the client terminal logs into the information processing platform, it sends a request to view the case, and there are many ways to trigger the request to view the case. For example, it can directly click the button of the case inquiry.
  • the disclosed patent case information also includes: patent examination process documents, process documents include one or more of the acceptance notice, correction notice, examination opinion notice, and examination response documents;
  • the patent case information includes: patent processing process documents, patent processing process documents include one or more of technical disclosure documents, case communication process documents, and case writing manuscripts.
  • the disclosed patent case information mentioned above also includes: process documents generated during the patent examination process.
  • Process documents include: one or more of the acceptance notice, correction notice, office action notice, and examination response document.
  • the information processing platform can directly receive official examination process documents (notification of acceptance, notice of correction, notice of examination opinion), and then store the received examination process documents in association with the corresponding case. In this way, customers can inquire and obtain all relevant documents on the information processing platform, which is convenient and quick.
  • establishing an agent information database includes: obtaining the basic information of the agent, the basic information includes: the name of the agent and the practice experience, the practice experience includes the name of the agency corresponding to the agent during the practice;
  • the database is connected, and the patent case information is stored in the third-party database.
  • the patent case information includes: the name of the agency and the name of the agent; all patent cases corresponding to the agent are extracted from the third-party database according to the name and practice experience of the agent Data; use the established patent case data processing model to process all the patent case data corresponding to the agent, and calculate the historical authorization rate of the agent; extract the preset number within the preset period from all the patent case data corresponding to the agent Quantity of patent cases to be evaluated; conduct technical analysis on the patent cases to be evaluated to determine the technical field to which the patent case to be evaluated belongs; send the patent cases to be evaluated to experts in the technical field for review according to the technical field, and receive the returned review results; obtain The historical service evaluation information corresponding to the agent; according to the agent’s historical authorization rate, evaluation results and historical service evaluation information, the professional level of the agent is determined according to the preset algorithm; all the patent case data corresponding to the agent are analyzed and processed, and the statistics are obtained The main applicant of the agent’s historical agency and the main technical field of the historical agency; according to the professional level of the agent
  • the establishment of a customer information database includes: obtaining basic information of the customer, the basic information includes: applicant information; establishing a connection with a third-party database, in which patent case information is stored; Capture all the patent case data corresponding to the applicant information in the tripartite database; use the preset patent analysis model to analyze all the patent case data corresponding to the applicant information, analyze and obtain the historical application information corresponding to the applicant information, historical application information Including: the technical field of the patent case of the historical application, the agency and the agent of the historical cooperation; the historical application information corresponding to the customer is stored in the customer information database.
  • FIG. 6 it is a sequence diagram of information interaction. It includes two parts. The first part is the sequence diagram for generating the agent's personal introduction information, and the second part is the sequence diagram for agent recommendation.
  • the agent registers through the agent terminal.
  • the agent’s basic information includes: the agent’s name and practice experience.
  • the information processing platform After receiving the basic information of the agent, the information processing platform sends a data request (carrying the name of the agent + practice experience) to the third-party database, and receives the patent case data returned by the third-party database. Then the information processing platform analyzes the patent case data, and calculates the historical authorization rate.
  • the information processing platform also extracts a preset number of patent cases to be evaluated within the preset period from the patent case data, and conducts technical analysis to determine The technical field of the patent case to be evaluated; after that, the patent case to be evaluated is sent to the corresponding expert for review, and the returned review result is received; in addition, the information processing platform obtains the historical service evaluation information corresponding to the agent from its own database; and then According to the agent's historical authorization rate, review results and historical service evaluation information, the professional level of the agent is determined; in addition, the information processing platform analyzes and processes all the patent case data corresponding to the agent, and obtains the statistics of the agent's historical representation The applicant and the technical field of the historical agent; finally, according to the professional level of the agent, the main applicant of the historical agent, and the main technical field of the historical agent, the personal introduction information corresponding to the agent is generated according to the preset template, And store the generated personal introduction information in the agent information database.
  • the second part the customer registers on the information processing platform through the client terminal, fills in the basic information of the customer (including at least the applicant information) when registering, and the information processing platform grabs the corresponding patent case data from the third-party database according to the applicant information, And use the preset patent analysis model to analyze all the patent case data corresponding to the applicant information, analyze and obtain the historical application information corresponding to the applicant information, the historical application information includes: the technology of the patent cases of the historical application Agencies and agents in the fields and historical cooperation are added to the customer information database; in addition, the information processing platform also obtains the agency demand information sent by the customer terminal, and obtains the corresponding historical evaluation information of the customer from its own database, and finally according to the agency demand Information, historical evaluation information and historical application information are matched with the agent information in the database to generate a matching agent list, which contains multiple matching agent information; the generated agent list is returned to to the client terminal.
  • the client terminal selects the intended agent, and then the information processing platform sends it to the agent terminal for confirmation. After the agent agree
  • an information processing device based on data interaction which includes:
  • An acquisition module 702 configured to acquire the basic information of the agent, the basic information including: identification information of the agent;
  • a connection module 704 configured to establish a connection with a third-party database, wherein patent case information is stored in the third-party database, and the patent case information includes: agent identification information;
  • the processing module 706 is used to use the established patent case data processing model to process all the patent case data corresponding to the agent, and calculate the historical authorization rate of the agent within a preset time;
  • An extraction module 708, configured to extract a preset number of patent cases to be evaluated within a preset number of years from all patent case data corresponding to the agent;
  • Review module 710 configured to send the patent case to be evaluated to experts in the technical field for review, and receive the returned review result
  • a determining module 712 configured to determine the professional level of the agent according to a preset algorithm according to the historical authorization rate of the agent and the review result;
  • the generating module 714 is configured to generate personal introduction information corresponding to the agent according to the professional level of the agent according to a preset template.
  • the above device further includes: an extracting module, configured to extract all patent case data corresponding to the attorney from the third-party database according to the attorney's name and the practice experience.
  • the basic information includes: the direction of a professional agent; the extraction module 708 is also used to screen out the candidate patent case data that matches the direction of the professional agent from all the patent case data corresponding to the agent; Patent cases to be evaluated are extracted from the candidate patent case data.
  • the processing module 706 is further configured to use the established patent case data processing model to extract the legal status in all the patent case data, and the legal status includes: one of authorized, pending, rejected, and withdrawn.
  • the patent case data processing model is used to determine the regular expression corresponding to the legal state, and extract the legal state of each case based on the regular expression; count the number of authorized cases whose legal state is in the authorized state, the legal state The number of rejected cases in the dismissed state and the number of withdrawn cases in the withdrawn state; the historical authorization rate is obtained by calculating the number of authorized cases, the number of rejected cases and the number of withdrawn cases.
  • the processing module 706 is also used to count the number of authorized cases, the number of rejected cases, and the number of withdrawn cases in the preset number of closed cases closest to the current time; according to the preset number closest to the current time The number of authorized cases among the closed cases, the number of rejected cases and the number of withdrawn cases are calculated to obtain the latest historical authorization rate.
  • the processing module 706 is also used to screen out candidate patent cases whose application dates are within a preset time period; count the number of authorized cases in the authorized state, the number of rejected cases in the rejected state among the candidate patent cases.
  • the number of withdrawn cases and the number of withdrawn cases in the withdrawn state calculated according to the number of authorized cases in the authorized state, the number of rejected cases in the rejected state and the number of withdrawn cases in the withdrawn state among the candidate patent cases
  • the historical authorization rate corresponding to the preset time period.
  • the determining module 712 is also used to obtain the agent's years of service and the total amount of historical case applications, and use the years of practice and the total amount of historical case applications as inputs to the weight determination model, so
  • the weight determination model is obtained by training and learning by using a deep neural network model; the first weight of the historical authorization rate corresponding to the agent output by the weight determination model, the second weight of the review result and the obtained
  • the third weight of the historical service evaluation information according to the historical authorization rate and the first weight, the evaluation result and the second weight, the historical service evaluation information and the third weight to determine the agent the professional level of the person.
  • the extraction module is also used to determine the target search condition according to the name of the agent and the practice experience, the target search condition includes the name of the agent, the name of the practicing agency and the corresponding practice time; according to The target search condition extracts all patent case data corresponding to the attorney from the third-party database.
  • the above-mentioned device further includes: an analysis module, configured to analyze and process all the patent case data corresponding to the agent, and obtain statistically the applicants represented by the agent and the technical fields represented historically.
  • an analysis module configured to analyze and process all the patent case data corresponding to the agent, and obtain statistically the applicants represented by the agent and the technical fields represented historically.
  • the analysis module is also used to capture the classification number information and abstract information of each patent case, analyze the abstract information, and extract technical keywords; use the classification number information and the technical keywords as the technical field
  • the input of the classification model is to determine the technical field category corresponding to the patent case, and use the determined technical field category as the technical field label of the patent case.
  • the technical field classification model includes: a first feature model, a second feature model and a classification model, the first feature model is used to determine the first feature vector corresponding to the classification number according to the classification number, and the second The characteristic model is used to determine the second characteristic vector corresponding to the technical keyword, and the classification model is used to determine the technical field category corresponding to the patent case according to the first characteristic vector and the second characteristic vector.
  • the basic information also includes: the qualification certificate number of the agent, and the above-mentioned device also includes:
  • the verification module is used to obtain the official practice experience of the agent according to the qualification certificate number of the agent; verify the basic information of the agent according to the official practice experience, and if the verification is successful, notify the connection
  • the module enters to establish a connection with the third-party database; if the verification fails, the result of the verification failure will be returned.
  • the statistical module is also used to extract the applicant information in each case, and count the number of cases corresponding to the same applicant information; determine the main applicant according to the number of cases corresponding to each applicant information, and the main applicant The applicant is the applicant with the highest number of cases; obtain the technical classification number and technical keywords of each case, and determine the technical field corresponding to each case according to the technical classification number and the technical keywords; count the same technology
  • the number of cases corresponding to the field, and the main technical field is determined according to the number of cases corresponding to the same technical field.
  • an information recommendation device based on data interaction including:
  • a connection module 804 configured to establish a connection with a third-party database, where patent case information is stored in the third-party database;
  • the analysis module 806 is configured to use a preset patent analysis model to analyze all patent case data corresponding to the applicant information, and analyze to obtain historical application information corresponding to the applicant information.
  • the historical application information includes: historical application The technical field of the patent case, the historical cooperation agency and agent;
  • the obtaining module is also used to obtain the client's agency demand information
  • a matching module 808, configured to match the agent demand information and the historical application information with the agent information in the database to generate a matching agent list, the agent list including a plurality of matching agent information;
  • Returning module 810 configured to return the generated agent list to the client terminal.
  • the above device further includes: a capture module, configured to capture all patent case data corresponding to the applicant information from the third-party database according to the applicant information.
  • the analysis module is also used to extract the technical classification number, technical keywords, agency name and agent corresponding to each patent case by using the preset patent analysis model; according to the technical classification number and The technical keywords determine the technical field corresponding to the patent case; extract the agency corresponding to each patent case, and count the number of first cases handled by each agency in history, according to the number of the first case, the historical cooperation of the The agency sorts; extracts the agent corresponding to each patent case, counts the number of second cases handled by each agent, and sorts the historical cooperation agents according to the number of second cases.
  • the acquisition module is also used to acquire the content of the agency demand input by the customer, perform semantic analysis on the content of the agency demand, and extract the semantic information of the content of the agency demand; determine the customer's content based on the semantic information Agent requirement information, the agent requirement information includes: at least one of field requirements, professional level requirements, professional requirements, and time limit requirements.
  • the agent information includes: professional level information, field information, professional information and time information; A candidate agent whose demand information matches; calculate the degree of matching with each of the candidate agents according to the agent demand information, the historical evaluation information and the historical application information; generate an agent list according to the degree of matching, the The agent information in the agent list is arranged according to the degree of matching.
  • the matching module is also used to determine field requirements, level requirements, professional requirements, and time limit requirements according to the agency requirement information, the historical evaluation information, and the historical application information; according to the field requirements and the Perform similarity calculation on the domain information corresponding to the candidate agent to obtain the domain matching degree corresponding to the domain requirement; perform correlation calculation according to the professional requirement and the professional information corresponding to the candidate agent to obtain the corresponding professional requirement professional matching degree; matching according to the level requirement and the level information corresponding to the candidate agent to obtain the level matching degree corresponding to the level requirement; matching according to the time limit requirement and the time information corresponding to the candidate agent , to obtain the time limit matching degree corresponding to the time limit requirement; determine the matching degree with each of the candidate agents according to the field matching degree, the professional matching degree, the level matching degree and the time limit matching degree.
  • the matching module is also used to obtain customer bias information from the agency demand information and the historical evaluation information
  • the bias information includes: the customer's demand for the field, the professional demand, the Level requirements, the degree of emphasis on the time limit requirements; the weight information is used as the input of the weight analysis model, and the field weight corresponding to the field matching degree output by the weight analysis model and the field weight corresponding to the professional matching degree are obtained.
  • the weight, the time limit requirement and the time limit weight are calculated to obtain the matching degree with each of the candidate agents.
  • the agent information includes: level information; the level information is determined by obtaining the basic information of the agent, the basic information of the agent includes: the name of the agent and the practice experience, The practice experience includes the name of the agency corresponding to the agent during the practice period; a connection is established with a third-party database, and the third-party database stores patent case information, and the patent case information includes: the name of the agency and the name of the agent name; extract all patent case data corresponding to the agent from the third-party database according to the name of the agent and the practice experience; use the established patent case data processing model to Process the patent case data, and calculate the historical authorization rate of the agent; extract the preset number of patent cases to be evaluated within the preset number of years from all the patent case data corresponding to the agent; Perform technical analysis on patent cases to determine the technical field to which the patent case to be evaluated belongs; send the patent case to be evaluated to experts in the technical field for review according to the technical field, and receive the returned review results; obtain all The historical service evaluation information corresponding to the agent; according to
  • the information recommendation based on data interaction further includes: a communication establishment module, configured to receive the intended agent selected from the agent list sent by the client terminal, and send the client's intention request to The intended agent terminal; receiving the intention result sent by the intended agent terminal, when the intention result is agreed, establishing an instant communication connection channel between the client terminal and the agent terminal, the instant The communication connection channel is used for case communication between the client and the agent.
  • a communication establishment module configured to receive the intended agent selected from the agent list sent by the client terminal, and send the client's intention request to The intended agent terminal; receiving the intention result sent by the intended agent terminal, when the intention result is agreed, establishing an instant communication connection channel between the client terminal and the agent terminal, the instant The communication connection channel is used for case communication between the client and the agent.
  • an information interaction device including:
  • the first establishing module 902 is used to establish an agent information database, the agent information database includes a plurality of agent information, and each of the agent information includes: the professional level of the agent;
  • the second establishment module 904 is used to establish a customer information database, which includes: a plurality of customer information, each of which includes: historical application information;
  • An acquisition module 906, configured to acquire agency demand information of target customers
  • the matching module 908 is configured to perform matching according to the agent demand information, the historical application information corresponding to the target customer, and the agent information in the agent information database to generate a matching agent list, which includes multiple matching agent information;
  • Returning module 910 configured to return the generated agent list to the client terminal.
  • the above-mentioned information interaction device further includes:
  • the receiving module is used to receive the intended agent selected from the agent list sent by the client terminal, and send the client's intention request to the intended agent terminal; receive the intention sent by the intended agent terminal result;
  • the communication establishment module is used to establish an instant messaging connection channel between the client terminal and the agent terminal when the intention result is consent, and the instant communication connection channel is used for carrying out the case between the client and the agent communicate;
  • the saving module is used to save the previous case communication process between the client and the agent as the case communication process file, and the case communication process file is stored in association with the corresponding patent case.
  • the first building module is further configured to interact with the third-party database, grab the published patent case information corresponding to the customer information from the third-party database, and the published patent case information Including: case review status; obtaining the patent case information to be processed corresponding to the customer information, the patent case information to be processed includes: case processing status; combining the published patent case information and the patent case information corresponding to the same customer
  • the case information to be processed is stored in the customer case information database corresponding to the same customer, and one customer corresponds to one customer case information database.
  • the above-mentioned information interaction device further includes:
  • the search module is used to receive the case viewing request sent by the client terminal, obtain the authority information corresponding to the client in response to the request, obtain the patent case information corresponding to the authority information from the client case information database, and return to the client terminal for display.
  • the disclosed patent case information also includes: patent examination process documents, which include one or more of the acceptance notice, correction notice, examination opinion notice, and examination response document.
  • the patent case information to be processed includes: patent processing process files, and the patent processing process files include one or more of technical disclosure documents, case communication process files, and case writing manuscripts.
  • the second establishment module is also used to acquire the basic information of the agent, the basic information includes: the name of the agent and practice experience, the practice history includes the name of the agency corresponding to the agent during practice;
  • a connection is established with a third-party database, and patent case information is stored in the third-party database, and the patent case information includes: the name of the agency and the name of the agent; Extract all the patent case data corresponding to the agent from the tripartite database; use the established patent case data processing model to process all the patent case data corresponding to the agent, and calculate the historical authorization rate of the agent; Extract a preset number of patent cases to be evaluated within a preset number of years from the data of all patent cases corresponding to the agent; perform technical analysis on the patent cases to be evaluated, and determine the technical field to which the patent cases to be evaluated belong ; According to the technical field, send the patent case to be evaluated to the corresponding expert in the technical field for review, and receive the returned review result; obtain the historical service evaluation information corresponding to the agent; according to the historical service evaluation
  • the main applicant and the main technical field of the historical agent according to the professional level of the agent, the main applicant of the historical agent and the main technical field of the historical agent, the personal introduction information corresponding to the agent is generated according to the preset template ; Store the personal introduction information corresponding to the agent in the agent information database.
  • the above-mentioned device further includes: a third establishing module, configured to obtain basic information of the client, the basic information including: applicant information; establishing a connection with a third-party database, in which patent information is stored. Case information; capture all patent case data corresponding to the applicant information from the third-party database according to the applicant information; use the preset patent analysis model to analyze all patent case data corresponding to the applicant information Analyze and obtain the historical application information corresponding to the applicant information, the historical application information includes: the technical field of the patent case of the historical application, the agency and the agent of the historical cooperation; the historical application information corresponding to the customer Stored in the customer information database.
  • a third establishing module configured to obtain basic information of the client, the basic information including: applicant information; establishing a connection with a third-party database, in which patent information is stored. Case information; capture all patent case data corresponding to the applicant information from the third-party database according to the applicant information; use the preset patent analysis model to analyze all patent case data corresponding to the applicant information Analyze
  • Figure 10 shows a diagram of the internal structure of a computer device in one embodiment.
  • the computer device may be a server, and the computer device includes a processor and a memory connected through a system bus.
  • the memory includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium of the computer device stores an operating system, and may also store a computer program.
  • the processor can realize the above-mentioned information processing method based on data interaction or the information processing method based on data interaction. Recommendation method or information interaction method.
  • a computer program may also be stored in the internal memory.
  • the processor may execute the above-mentioned information processing method based on data interaction, information recommendation method or information interaction method based on data interaction.
  • the structure shown in Figure 10 is only a block diagram of a part of the structure related to the solution of this application, and does not constitute a limitation to the computer equipment on which the solution of this application is applied.
  • the specific computer equipment can be More or fewer components than shown in the figures may be included, or some components may be combined, or have a different arrangement of components.
  • a computer-readable storage medium storing a computer program.
  • the processor executes the above-mentioned information processing method based on data interaction or information recommendation method or information interaction method based on data interaction. step.
  • a computer device including a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the above information processing method based on data interaction or the information processing method based on data interaction Steps of an information recommendation method or an information interaction method.
  • Nonvolatile memory can include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory can include random access memory (RAM) or external cache memory.
  • RAM is available in many forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Chain Synchlink DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
  • SRAM Static RAM
  • DRAM Dynamic RAM
  • SDRAM Synchronous DRAM
  • DDRSDRAM Double Data Rate SDRAM
  • ESDRAM Enhanced SDRAM
  • SLDRAM Synchronous Chain Synchlink DRAM
  • Rambus direct RAM
  • DRAM direct memory bus dynamic RAM
  • RDRAM memory bus dynamic RAM

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Data Mining & Analysis (AREA)
  • Technology Law (AREA)
  • Development Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Economics (AREA)
  • Tourism & Hospitality (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Operations Research (AREA)
  • Human Resources & Organizations (AREA)
  • Primary Health Care (AREA)
  • Game Theory and Decision Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本申请提供了一种基于数据交互的信息推荐方法，包括：获取客户的基础信息，与第三方数据库建立连接，对所述申请人信息对应的所有专利案件数据进行分析得到所述申请人信息对应的历史申请信息，获取客户的代理需求信息；根据所述代理需求信息和所述历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息，将生成的所述代理人列表返回给客户终端。该方法实现了为客户准确地进行代理人推荐。此外，还提出了一种基于数据交互的信息推荐装置、设备及存储介质。

Description

基于数据交互的信息推荐方法、装置、设备及存储介质 技术领域
本申请涉及信息处理技术领域，特别涉及为一种基于数据交互的信息推荐方法、装置、设备及存储介质。
背景技术
随着科技的蓬勃发展，知识产权越来越受到企业的重视，专利作为一种无形资产，已经成为了企业的一种重要的竞争资源。专利文件是一种专业的法律文件，申请人往往需要委托专利代理机构进行专利申请，专利文件的撰写是由代理机构中的专利代理师来完成的。
然而，目前专利代理行业内存在着信息严重不对称的问题，如果申请人(即客户)想要了解专利代理师的专业水平，只能通过代理机构提供的关于专利代理师的介绍信息来了解，这种了解专利代理师的专业水平的渠道不仅效率低，而且也无法保障代理机构提供的专利代理师的资料是真实可信的。
即使能够获取到准确的专利代理师信息，但是面对大量的专利代理师信息，想要从中选出合适的专利代理人，对于客户来说不仅难度大，而且效率低。为此，行业内提出了一些自动匹配推荐专利代理师的方法，但是目前行业内进行匹配推荐的方式都只是适用于代理机构内部的推荐，其推荐过程中依据的信息有限，不能准确地推荐出真正符合客户需求的专利代理师。
因此，亟需要提出一种适用范围广且更加准确地匹配出专利代理师的方法。
发明内容
本申请旨在解决上述推荐不准确的问题，提出了一种基于数据交互的信息推荐方法、装置、设备及存储介质，该方法能够更加准确地匹配出专利代理师。
为了解决上述推荐不准确的问题，第一方面提出了一种基于数据交互的信息推荐方法，应用于信息处理平台，包括：
获取客户的基础信息，所述客户的基础信息包括：申请人信息；
与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息；
采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息；
获取客户的代理需求信息；
根据所述代理需求信息和所述历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息；
将生成的所述代理人列表返回给客户终端。
为了解决上述推荐不准确的问题，第二方面提出了一种基于数据交互的信息推荐装置，包括：
获取模块，用于获取客户的基础信息，所述客户的基础信息包括：申请人信息；
连接模块，用于与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息；
分析模块，用于采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息；
所述获取模块还用于获取客户的代理需求信息；
匹配模块，用于根据所述代理需求信息和所述历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息；
返回模块，用于将生成的所述代理人列表返回给客户终端。
为了解决上述推荐不准确的问题，第三方面提供了一种计算机可读存储介质，存储有计算机程序，所述计算机程序被处理器执行时，使得所述处理器执行以下步骤：
获取客户的基础信息，所述客户的基础信息包括：申请人信息；
与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息；
采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息；
获取客户的代理需求信息；
根据所述代理需求信息和所述历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息；
将生成的所述代理人列表返回给客户终端。
为了解决上述推荐不准确的问题，第四方面提供了一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述计算机程序被所述处理器执行时，使得所述处理器执行以下步骤：
获取客户的基础信息，所述客户的基础信息包括：申请人信息；
与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息；
采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息；
获取客户的代理需求信息；
根据所述代理需求信息和所述历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息；
将生成的所述代理人列表返回给客户终端。
上述基于数据交互的信息推荐方法、装置、设备及存储介质，首先，获取客户信息(申请人信息)，与第三方数据库建立连接，然后对申请人信息对应的所有专利案件数据进行分析提取出历史申请信息，之后还需要获取客户的代理需求信息，进而根据历史申请信息和代理需求信息进行代理人信息匹配，生成匹配的代理人列表，并将生成的代理人列表返回给客户终端。该基于数据交互的信息推荐方法，通过多方面的信息匹配，实现了为客户准确推荐代理人。另外，该推荐方法不局限于某个代理机构内部的推荐，适用于针对所有代理人的推荐，适用范围广泛。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为一个实施例中基于数据交互的信息处理方法的流程图；
图2为一个实施例中确定代理人的专业级别的方法流程图；
图3为一个实施例中基于数据交互的信息推荐方法的流程图；
图4为一个实施例中计算与每个候选代理人的匹配度的方法流程图；
图5为一个实施例中信息交互方法的流程图；
图6为一个实施例中信息交互过程的时序图；
图7为一个实施例中基于数据交互的信息处理装置的结构框图；
图8为一个实施例中基于数据交互的信息推荐装置的结构框图；
图9为一个实施例中信息交互装置的结构框图；
图10为一个实施例中计算机设备的内部结构框图。
具体实施方式
应当理解，此处所描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
下面将结合本申请的实施例中的附图，对本申请的实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请的一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
需要说明的是，本申请的说明书和权利要求书及上述附图中的术语“包括”、“包含”和“具有”以及它们任何变形，意图在于覆盖不排他的包含。例如包含了一系列步骤或单元的过程、方法、系统、产品或设备没有限定于已列出的步骤或单元，而是可选地还包括没有列出的步骤或单元，或可选地还包括对于这些过程、方法、产品或设备固有的其他步骤或单元。在本申请的权利要求书、说明书以及说明书附图中的术语，诸如“第一”和“第二”等之类的关系术语仅仅用来将一个实体/操作/对象与另一个实体/操作/对象区分开来，而不一定要求或者暗示这些实体/操作/对象之间存在任何这种实际的关系或者顺序。
在本文中提及“实施例”意味着，结合实施例描述的特定特征、结构或特性可以包含在本申请的至少一个实施例中。在说明书中的各个位置出现该短语并不一定均是指相同的实施例，也不是与其他实施例互斥的独立的或备选的实施例。本领域技术人员显式地和隐式地理解的是，本文所描述的实施例可以与其他实施例相结合。
如图1所示，为了实现真实地展现专利代理人的个人介绍信息，提出了一 种基于数据交互的信息处理方法，应用于信息处理平台，包括：
步骤102，获取代理人的基础信息，基础信息包括：代理人身份识别信息。
其中，这里的代理人是指从事专利代理工作的人员，又称“专利代理人”或“专利代理师”。代理人基础信息中包括代理人身份识别信息，代理人身份识别信息适用于确定代理人身份的信息，在一个实施例中，可以直接采用代理人的姓名作为代理人身份的信息，也可以采用代理人的代理人资格证号作为代理人身份识别信息。在另一个实施例中，为了更加准确的确定代理人身份识别信息，采用代理人姓名+执业经历作为代理人身份识别信息，其中，执业经历中包括代理人在各个执业期间对应的代理机构名称。另外，代理人的基础信息中除了包括代理人身份识别信息外，还可以包括其他个人信息，比如，性别信息，学历信息、专业信息、代理人资格证信息、代理人执业证信息、代理人执业年限以及专业代理方向等。执业经历包括：代理人在何时何代理机构进行执业，比如，代理人李三，在2012.1-2018.7期间在A代理机构执业，在2018.8-2021.4期间在B代理机构执业。代理人的专业代理方向。目前行业内主要分为软通、电子、机械和生化四个专业代理方向。
代理人基础信息的获取方式可以有多种，在一个实施例中，代理人的基础信息可以直接从代理人的注册信息中获取，代理人需要在信息处理平台进行实名注册，注册时需要填入个人的相关信息。在另一个实施例中，代理人的基础信息可以通过与第三方平台进行交互，抓取代理人的基础信息，比如，从“中华全国代理人协会网站”查找到代理人的执业经历，从“学信网”可以查找到代理人的学历信息、专业信息等。
步骤104，与第三方数据库建立连接，第三方数据库中存储有专利案件信息，专利案件信息中包括：代理人身份识别信息。
其中，第三方数据库是指存储有专利案件信息的数据库。第三方数据库可以是国家知识产权局的专利检索库，当然也可以是其它的专利检索库，比如，智慧芽、佰腾、incopat等。对于由代理所提交的案件，专利案件信息中都包含有代理人身份识别信息，比如，每个专利案件信息中都包含代理机构名称和代理人姓名。
步骤106，采用建立的专利案件数据处理模型对代理人对应的所有专利案件数据进行处理，统计出代理人在预设时间内的历史授权率。
其中，专利案件数据处理模型用于对代理人对应的所有专利案件数据进行批量处理，通过该专利案件数据处理模型可以提取出每件专利案件的法律状态，法律状态包括：授权、驳回、审中、撤回四种状态。其中，撤回分为主动撤回和视为撤回。预设时间可以自定义设置，比如，将代理人从业以来的时间作为预设时间，也可以将最近五年作为预设时间。
在一个实施例中，历史授权率可以通过计算授权案件数量与已结案案件数量的比值计算得到，其中，已结案案件包括：授权案件、驳回案件和撤回案件。比如，授权案件数量为120，已结案案件数量为200，那么计算得到的历史授权率为120/200＝60％。
在一个实施例中，授权率的计算只基于发明专利案件进行统计计算，由于实用新型和外观专利不涉及实质审查，所以实用新型和外观专利的授权率不能准确地反映出代理人的专业水平。具体地，专利案件数据处理模型首先可以通过案件类型提取出发明专利，然后再提取出每件发明专利的法律状态，基于每件发明专利的法律状态计算得到代理人的历史授权率。通过基于发明专利进行授权率的统计有利于准确地对代理人的专业水平进行评价。
在另一个实施例中，为了更能够反映出代理人的当前专业水平，可以选取出距离当前时间最近的预设数量(比如，100件)的已结案案件，根据已结案案件中的授权数量、驳回数量和撤回数量计算得到最能够反映代理人当前专业水平的历史授权率。
步骤108，从代理人对应的所有专利案件数据中抽取出预设年限内的预设数量的待评价专利案件。
其中，待评价专利案件是随机抽取出来的用于由专家对代理人的撰写质量进行评价的案件。由于代理人的专业水平一般来说是随着经验的积累而逐步提高的，所以为了真实反映出代理人当前的专业水平，可以从预设年限内(比如，近三年)的专利案件数据中抽取出预设数量(比如，3篇)的专利案件作为待评价专利案件，从而有利于实现对该专利代理人的当前专业水平进行准确的评价。
步骤110，将待评价专利案件发送给对应的专家进行评审，接收返回的评审结果。
其中，为了更准确地对代理人的撰写质量进行评价，需要将待评价专利案 件发送给相应技术领域的专家进行评审，并接收返回的评审结果，评审结果可以是分数，也可以是等级。当有多篇待评价专利案件时，需要综合多篇待评价专利案件对应的评审结果。比如，当有三篇待评价专利案件时，可以取三件待评价专利案件的平均分。
在另一个实施例中，为了准确地匹配出评审专家，需要对待评价专利案件进行技术分析，确定待评价专利案件所属的技术领域，然后将待评价专利案件发送给技术领域对应的专家进行评审。其中，技术领域的划分可以根据具体的技术进行划分，具体地，技术领域的确定可以根据技术分类号进行确定，不同的技术分类号本身就代表了所属的技术领域。比如，G01表示测量、测试领域。G02表示光学。在另一个实施例中，技术领域的确定可以同时根据技术分类号和技术关键字来确定，技术分类号+技术关键字的组合能够得到更加细分的技术领域。比如，技术分类号表示程序控制，技术关键字为温度控制，那么确定的技术领域为：利用软件的温度控制。
步骤112，根据代理人的历史授权率和评审结果按照预设算法确定代理人的专业级别。
其中，代理人的历史授权率能够准确地反映出专业水平，评审结果能够准确地反映出案件的撰写质量，综合两者的结果可以准确地实现对代理人的专业分级。在一个实施例中，可以通过加权求和的方式将两者进行综合。
在一个具体的实施例中，获取代理人的从业年限和总的历史案件申请量，将从业年限和总的历史案件申请量作为权重确定模型的输入，权重确定模型是采用深度神经网络模型进行训练学习得到的。获取权重确定模型输出的代理人对应的历史授权率的第一权重和评审结果的第二权重；根据历史授权率和第一权重，评审结果和第二权重确定代理人的专业级别。
在另一个实施例中，还可以获取获取代理人对应的历史服务评价信息。
其中，历史服务评价信息是指代理人历史服务的客户对该代理人的专业和服务的评价信息，历史服务评价信息具体可以是分数，也可以是等级，比如，等级可以分为非常满意、满意、一般、不满意。代理人的历史服务评价信息可以真实地反映出代理人的服务质量，包括：服务态度和专业认可度。然后根据代理人的历史授权率、评审结果和历史评价信息按照预设算法确定代理人的专业级别。即综合三者的结果进行等级划分，该方法可以更加准确地实现对代理 人的专业分级。
需要说明的是：对于在信息处理平台还没有处理案件的代理人来说，还没有历史服务评价信息，即获取到的历史服务评价信息可能是空。当历史服务评价信息为空时，可以将历史评价信息设置为默认值，也可以将该项的权重设置为0，而加重其它两项的权重。在一个实施例中，专业级别的等级划分可以根据需要自定义设置，比如，专业级别可以分为五个等级，分别为AAA、AA+、AA、AA-、A。当然，也可以设置6个等级，此外，等级的表现形式也可以自定义设置，比如，也可以通过点亮几颗星星的方式作为表现形式。
步骤114，根据代理人的专业级别按照预设模板生成代理人对应的个人介绍信息。
其中，代理人的个人介绍信息中包括有代理人的专业级别，这样，可以更加直观地反映出代理人专业水平。另外，为了更全面地展示代理人的专业水平，生成的专利代理人的个人介绍信息中不仅包括能够反映代理人专业水平的专业级别，还包括：代理人历史代理的主要申请人和历史代理的主要技术领域。当然还可以包括其它信息，比如，包括代理人的专业代理方向。目前行业内主要分为软通、电子、机械和生化四个专业代理方向，当然也可以根据实际情况的变更自行调整，比如，进一步细化相应的代理方向。
在一个实施例中，代理人历史代理的主要申请人和历史代理的主要技术领域是通过如下方式得到的：对代理人对应的所有专利案件数据进行分析处理，统计得到代理人历史代理的申请人和历史代理的技术领域。
其中，当历史代理的申请人和技术领域比较多时，可以从中选择出主要申请人和主要技术领域。主要申请人是指将历史代理的各个申请人的案件数量从多到少进行排序，选出前预设数量的申请人作为主要申请人。比如，历史代理A公司10件专利，B公司20件专利，C公司34件专利，D公司5件专利，E公司12件专利，那么按照专利案件数量将申请人进行排序，分别是C公司、B公司，E公司、A公司和D公司，然后选出前三个申请人作为主要申请人，分别为C公司、B公司和E公司。主要技术领域是指将历史代理的各个技术领域的案件数量进行排序，选出前预设数量(比如，前三个)的技术领域作为主要技术领域。通过选取主要申请人和主要技术领域便于后续更加有重点地进行代理人信息的展示。
在一个实施例中，上述方法包括：将生成的个人介绍信息发送到终端进行展示。即终端登录到信息处理平台后，便可以查看代理人的个人介绍信息。解决了行业内无法准确快速获取代理人的个人介绍信息的问题。
上述基于数据交互的信息处理方法、装置、设备及存储介质，首先根据代理人基础信息通过与第三方数据库进行交互获取到代理人历史所代理的所有专利案件数据，然后对代理人对应的所有专利案件数据进行分析统计出代理人的历史授权率，进一步的，随机抽取预设年限内预设数量的待评价专利案件，将待评价专利案件发送给相应技术领域的专家进行评审，接收返回的评审结果，最后根据代理人对应的历史授权率、评审结果计算得到代理人的专业级别。
该方法中基于与第三方数据库交互可以抓取到真实的代理人历史代理的专利案件数据，基于该真实的专利案件数据可以得到该代理人真实的历史授权率，且通过抽取真实的专利案件进行专家评审，得到真实的评审结果。最后，通过综合代理人对应的历史授权率、评审结果来确定代理人的专业级别，该专业级别的确定真实可信。另外，然后根据代理人的专业级别按照预设模板生成代理人的个人介绍信息，该生成的代理人的个人介绍信息不但真实可信，而且全面，便于申请人根据代理人的个人介绍信息可以全面且准确地了解代理人的专业水平。另外，该代理人的个人介绍信息是对外开放的，申请人在终端上可以快速获取到，且可供申请人查看的是所有专利代理师的个人介绍信息，而不局限于只能查看某个代理机构的专利代理师。
在一个实施例中，所述代理人身份识别信息包括:代理人姓名和执行经历，所述执业经历中包括代理人在执业期间对应的代理机构名称；在所述采用建立的专利案件数据处理模型对所述代理人对应的所有专利案件数据进行处理之前，所述方法还包括：根据所述代理人姓名和所述执业经历从所述第三方数据库中提取出所述代理人对应的所有专利案件数据。
其中，由于代理人中很可能会出现同名同姓的情况，所以仅仅根据代理人姓名进行专利案件数据的提取很可能不准确，故，为了保障提取到准确的专利案件数据，本实施例中提出同时根据代理人姓名和代理人的执业经历从第三方数据库提取与代理人对应的所有专利案件数据。具体地，代理人的执业经历包括何时在何代理机构执业，所以需要根据代理人的执业经历限定提取条件，以 “姓名+时间段+代理机构”作为提取条件从第三方数据库中提取出与该代理人对应的所有专利案件数据。
在一个实施例中，基础信息包括：专业代理方向；从代理人对应的所有专利案件数据中抽取出预设年限内的预设数量的待评价专利案件，包括：从代理人对应的所有专利案件数据中筛选出与专业代理方向匹配的候选专利案件数据；从候选专利案件数据中抽取出待评价专利案件。
其中，专业代理方向是指代理人擅长的代理方向，主要分为软通(软件通信)、电子、机械和生化(生物和化学)四个专业代理方向。为了能够更加准确地反映代理人在其专业代理方向上的专业水平，在抽取待评价专利案件时，抽取与其专业代理方向匹配的专利案件。首先，筛选出与专业代理方向匹配的候选专利案件数据，然后再从候选专利案件中随机抽取几个待评价专利。通过与专业代理方向匹配，能够挑选出更加反映出代理人专业水平的待评价专利，从而有利于后续得到更加准确的评审结果。
在一个实施例中，采用建立的专利案件数据处理模型对代理人对应的所有专利案件数据进行处理，统计出代理人的历史授权率，包括：采用建立的专利案件数据处理模型提取所有专利案件数据中法律状态，法律状态包括：授权、审中、驳回和撤回中的一种，专利案件数据处理模型用于确定法律状态对应的正则表达式，基于正则表达式进行各个案件的法律状态的提取；统计法律状态处于授权状态的授权案件数量、法律状态处于驳回状态的驳回案件数量和法律状态处于撤回状态的撤回案件数量；根据授权案件数量、驳回案件数量和撤回案件数量计算得到历史授权率。
其中，专利案件数据处理模型在提取案件法律状态时采用的是正则表达式的方式，正则表达式是一种描述字符串匹配的模式。通过确定法律状态对应的正则表达式从各个案件中提取相应的法律状态，根据提取到的授权案件数量、驳回案件数量和撤回案件数量计算得到历史授权率，历史授权率＝授权案件数量/(授权案件数量+驳回案件数量+撤回案件数量)。
在一个实施例中，统计法律状态处于授权状态的授权案件数量、法律状态处于驳回状态的驳回案件数量和法律状态处于撤回状态的撤回案件数量，包括：统计距离当前时间最近的预设数量的已结案案件中的授权案件数量、驳回案件数量和撤回案件数量；根据授权案件数量、驳回案件数量和撤回案件数量计算 得到历史授权率，包括：根据距离当前时间最近的预设数量的已结案案件中的授权案件数量、驳回案件数量和撤回案件数量计算得到最新的历史授权率。
其中，已结案案件包括授权案件、驳回案件和撤回案件。为了更好地反映出代理人当前的专业水平，采用获取距离当前时间最近的预设数量的(比如，100件)已结案案件中的授权率作为历史授权率。该历史授权率是动态计算的，随着时间的推移，不断更新历史授权率，从而可以及时反映出代理人的专业水平。
在一个实施例中，统计法律状态处于授权状态的授权案件数量、法律状态处于驳回状态的驳回案件数量和法律状态处于撤回状态的撤回案件数量，包括：筛选出申请日在预设时间段内的候选专利案件；统计候选专利案件中处于授权状态的授权案件数量、处于驳回状态的驳回案件数量和处于撤回状态的撤回案件数量；根据授权案件数量、驳回案件数量和撤回案件数量计算得到历史授权率，包括：根据候选专利案件中处于授权状态的授权案件数量、处于驳回状态的驳回案件数量和处于撤回状态的撤回案件数量计算得到预设时间段对应的历史授权率。
其中，为了能够更加准确地反映代理人的当前专业水平，在计算历史授权率时，统计预设时间段内的案件的授权率。预设时间段可以是指申请日在具体某一年的时间段，由于专利审查的周期比较长，一般需要一年以上，甚至3年，所以可以先筛选出申请日近三年的案件，然后统计申请日在近三年的候选专利案件中的授权情况，包括：授权案件数量、驳回案件数量和撤回案件数量，从而能够反映出代理人在近三年的专业水准，这样计算得到的授权率更加接近于代理人的当前专业水平。
如图2所示，在一个实施例中，根据代理人的历史授权率、评审结果和历史服务评价信息按照预设算法确定代理人的专业级别，包括：
步骤202，获取代理人的从业年限和总的历史案件申请量，将从业年限和总的历史案件申请量作为权重确定模型的输入，权重确定模型是采用深度神经网络模型进行训练学习得到的。
其中，为了能够更加准确地反映出代理人的真实专业水平，需要根据代理人的从业年限和总的历史案件申请量来调整上述历史授权率、评审结果和历史服务评价信息三者的权重。具体的调整是根据训练得到的权重确定模型进行调 整的。
权重确定模块是基于深度神经网络模型进行训练学习得到的，训练权重确定模型采用有监督训练方式，需要构建训练数据，训练数据为代理人的从业年限和总的历史案件申请量，然后将相应的人工标注的各个权重作为标签进行训练，然后基于设置的损失函数采用梯度下降法不断地调整权重确定模型中的参数，直到模型达到收敛条件。
步骤204，获取权重确定模型输出的代理人对应的历史授权率的第一权重，评审结果的第二权重和历史服务评价信息的第三权重。
其中，获取权重确定模型输出的第一权重、第二权重和第三权重，第一权重反映了历史授权率对于代理人专业等级的影响程度，第二权重反映了评审结果对于代理人专业等级的影响程度，第三权重反映了历史评价信息对于代理人专业等级的影响程度。一般来说，从业年限越短的代理人，评审结果的权重相对较大，历史授权率和历史服务评价信息的权重较小，这是因为专利审查周期长，从业年限短的代理人的历史授权率往往不高，或者还没有相关数据，同样地，对应的了历史服务评价信息也会较少，所以参考意义不大。随着从业年限增长，代理的总的案件数量增多，历史授权率和历史服务评价信息的权重相对会增加。
步骤206，根据历史授权率和第一权重，评审结果和第二权重、历史服务评价信息和第三权重确定代理人的专业级别。
其中，在确定了历史授权率对应的第一权重、评审结果对应的第二权重和历史服务评价信息对应的第三权重后，采用加权求和即可确定代理人的专业级别。具体地，在具体计算时，历史授权率、评审结果和历史服务评价信息都可以换算成分数，计算的得到综合分数后，根据综合分数确定代理人的专业级别。比如，分为五个级别。通过采用权重确定模型输出历史授权率、评审结果和历史服务评价信息的权重，然后通过加权求和的方式计算得到代理人的专业级别，从而能够更加真实地反映出代理人的专业水平。
在一个实施例中，根据代理人姓名和执业经历从第三方数据库中提取出与代理人对应的所有专利案件数据，包括：根据代理人姓名和执业经历确定目标搜索条件，目标搜索条件中包括代理人姓名、执业的代理机构名称以及对应的执业时间；根据目标搜索条件从第三方数据库中提取出代理人对应的所有专利 案件数据。
其中，为了更准确地抓取到与代理人对应的所有专利案件数据，需要准确的目标搜索条件，目标搜索条件需要同时包括代理人姓名+执业的代理机构名称+相应的执业时间。通过设置该目标搜索条件可以尽可能地排除掉同名同姓所带来的干扰。从而更有利于准确地获取与该代理人对应的所有专利案件数据。
在一个实施例中，对待评价专利案件进行技术分析，确定待评价专利案件所属的技术领域，包括：获取待评价专利案件对应的技术分类号和领域关键字；根据技术分类号和领域关键字确定待评价专利所属的技术领域。
其中，领域关键字是指表示该技术所应用的领域，技术分类号用于确定技术分类，根据技术分类号和领域关键字可以确定具体的技术领域，比如，根据技术分类号确定的是温度控制技术，领域关键字为：电子烟，那么就确定待评价专利属于电子烟的温度控制。
在一个实施例中，对代理人对应的所有专利案件数据进行分析处理，统计得到代理人历史代理的申请人和历史代理的技术领域，包括：抓取每件专利案件的分类号信息和摘要信息，对摘要信息进行分析，提取出技术关键词；将分类号信息和技术关键词作为技术领域分类模型的输入，确定专利案件对应的技术领域类别，将确定的技术领域类别作为专利案件的技术领域标签。
其中，分类号信息可以具体到小组，比如，G06K9/00。摘要信息是指专利文件中的摘要内容。对摘要信息进行分析提取出技术关键词，技术关键词的提取可以采用基于语义的提取方式，即通过对摘要信息进行语义分析提取出技术关键词。然后采用技术领域分类模型确定相应的专利案件的技术领域类别，将该技术领域类别作为该专利案件的技术标签进行存储，从而不但方便申请人查看到代理人历史代理的技术领域信息，而且便于后续进行专利技术领域的匹配推荐。
在一个实施例中，技术领域分类模型包括：第一特征模型、第二特征模型和分类模型，第一特征模型用于根据分类号确定分类号对应的第一特征向量，第二特征模型用于确定技术关键词对应的第二特征向量，分类模型用于根据第一特征向量和第二特征向量确定专利案件对应的技术领域类别。
其中，第一特征模型用于将分类号转换为第一特征向量，第二特征模型用于将技术关键词转换为第二特征向量，然后将第一特征向量和第二特征向量进 行组合，将组合后的特征向量作为分类模型的输入，分类类型用于根据第一特征向量和第二特征向量中包含的信息对专利案件进行技术领域分类。
构建训练集是模型训练最重要的一环，传统的构建训练集往往需要耗费大量的人力物力进行训练数据的标注，为了提高标注的效率，提出了一种快速标注的数据的方法。根据技术分类号和技术关键词在第三方数据库中查找与技术分类号和技术关键词对应的专利案件，将查找到的专利案件统一标注为与技术分类号和技术关键词对应的技术领域类别，从而不但快速获取到了训练数据，而且实现了对训练数据的快速标注。
在一个实施例中，基础信息还包括：代理人资格证号，在获取代理人的基础信息之后，还包括：根据代理人资格证号获取代理人的官方执业经历；根据官方执业经历对代理人的基础信息进行校验，若校验成功，则进入与第三方数据库建立连接的步骤；若校验失败，则返回校验失败的结果。
其中，为了保障获取到的代理人基础信息是准确的，在获取到基础信息后，还需要对获取得到的信息进行校验，校验的方式是：根据代理人资格证号查找该代理人的官方执业经历，然后根据官方执业经历对代理人的基础信息进行校验。具体地，可以通过与“中华全国代理人协会网站”进行交互来获取代理人的官方执业经历。然后与注册信息中填写的执业经历进行比对，从而确定填写的信息是否准确。
在一个实施例中，对代理人对应的所有专利案件数据进行分析处理，统计得到代理人历史代理的申请人和历史代理的技术领域，包括：提取每一件案子中的申请人信息，统计同一申请人信息对应的案件数量；根据每个申请人信息对应的案件数量确定主要申请人，主要申请人为案件数量排名靠前的申请人；获取每一件案子的技术分类号和技术关键字，根据技术分类号和技术关键字确定每一件案子对应的技术领域；统计同一技术领域对应的案件数量，根据同一技术领域对应的案件数量确定主要技术领域。
其中，为了更详细地展示代理人的相关信息，统计代理人之前代理的主要申请人，以及历史代理的主要技术领域。主要申请人是指历史服务的主要客户，主要技术领域是指历史服务的主要技术领域。比如，代理人李三，主要服务的客户有A公司、B公司、C公司、D公司等，主要技术领域为：I领域、II领域、III领域等。这样便于后续客户可以根据代理人的个人介绍信息对该代理人有 更加准确的了解，从而选择最适合自己的代理人。
当面对大量的专利代理师信息时，如何从中快速选择出合适的代理人对于客户也是一个难题，如果一个个地去浏览各个专利代理人的个人介绍信息，无疑将会耗时耗力，且对于非专业人士，挑选出来的代理人也未必是最合适的。基于此，提出了一种基于数据交互的信息推荐方法，该方法可以快速准确地匹配出专利代理人，且该方法适用范围广泛。
如图3所示，提出了一种基于数据交互的信息推荐方法，应用于信息处理平台，包括：
步骤302，获取客户的基础信息，客户的基础信息包括：申请人信息。
其中，申请人信息可以是个人姓名，也可以是企业名称。对于企业客户，申请人信息是指企业的全称。一个客户账号可以关联一个或多个申请人信息。此外，客户的基础信息还可以包括：企业资质信息、企业的法人信息等。
步骤304，与第三方数据库建立连接，第三方数据库中存储有专利案件信息。
其中，第三方数据库是指存储有专利案件信息的数据库。第三方数据库可以是国家知识产权局的专利检索库，当然也可以是其它的专利检索库，比如，智慧芽、佰腾、incopat等。对于由代理所提交的案件，专利案件信息中都包含有申请人信息。
步骤306，采用预设的专利分析模型对申请人信息对应的所有专利案件数据进行分析，分析得到申请人信息对应的历史申请信息。
其中，专利分析模型用于对抓取得到的所有专利案件数据进行分析，得到申请人的历史申请情况。历史申请信息包括：历史申请的专利案件的技术领域、历史合作的代理机构、历史合作的代理人中的至少一个。具体地，可以通过分类号分析出专利案件的技术领域，然后从专利案件数据中提取出历史合作的代理机构以及代理人等信息。通过获取申请人的历史申请的专利案件数据有利于分析得到客户的实际需求。在另一个实施例中，通过分类号和从摘要信息中提取到的技术关键词确定技术领域。
步骤308，获取客户的代理需求信息。
其中，代理需求信息是指客户对代理人的要求信息。代理需求信息包括但 不限于：领域需求、级别需求、专业需求、代理机构需求、时限需求、位置需求等中的一个或多个。领域需求是指对代理人的专业代理方向的要求，专业代理方向可以分为软通、电子、机械和生化四个方向。级别需求是指对代理人的级别要求，代理人的级别越高，说明该代理人的撰写质量越好，当然相应的代理费用也会越高。专业需求是指对代理人所学专业的要求，比如，对于LED领域的客户，其往往要求代理人懂得LED相关的知识，所以会优先选择光电领域的代理人。代理机构需求反应的是客户对某个代理机构的信任度，如果客户之前和某个代理机构合作的比较愉快，可以优先选择该代理机构下的代理人进行合作。时限需求是指对代理人返稿时间的限定，比如，对于比较紧急的案件，客户要求在短时间内返稿。位置需求是指对代理人所在位置的要求，比如，有些客户想要当面和代理人进行案件沟通，会优先选择距离自己比较近的代理人。
步骤310，根据代理需求信息和历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，代理人列表中包含多个匹配的代理人信息。
其中，综合代理需求信息和历史申请信息，然后与数据库中的代理人信息进行匹配，生成代理人列表，在代理人列表中可以按照匹配度从高到低进行排序，从而便于后续客户进行选择。这里的代理人是指从事专利代理工作的人员，又称“专利代理人”或“专利代理师”。
在一个实施例中，代理人列表中也可以分多个维度对代理人排序，比如，可以分为专业维度，即专业匹配度最高的，级别维度，即级别匹配度最高的，以及综合维度，即综合了多个指标的维度排名，客户可以根据偏重来选择排序维度，从而有利于为客户提供更加灵活的选择。
在一个实施例中，还需要获取客户对应的历史评价信息，根据代理需求信息、历史申请信息和历史评价信息与数据库中的代理人进行匹配，生成代理人列表。其中，历史评价信息是指客户对历史代理的代理人的评价信息。历史评价信息能够反映出客户对历史代理的代理人的认可度，如果认可度高，那么后续可以优先推荐该代理人，如果认可度低，则后续不再推荐该代理人。
步骤312，将生成的代理人列表返回给客户终端。
其中，将代理人列表返回给客户终端，客户根据代理人列表进行选择，大大提高了选择的准确度。
上述基于数据交互的信息推荐方法、装置、设备及存储介质，首先，获取 客户信息(申请人信息)，与第三方数据库建立连接，然后对申请人信息对应的所有专利案件数据进行分析提取出历史申请信息，之后还需要获取客户的代理需求信息，进而根据历史申请信息和代理需求信息进行代理人信息匹配，生成匹配的代理人列表，并将生成的代理人列表返回给客户终端。该基于数据交互的信息推荐方法，通过多方面的信息匹配，实现了为客户准确推荐代理人。另外，该推荐方法不局限于某个代理机构内部的推荐，适用于针对所有代理人的推荐，适用范围广泛。
在一个实施例中，所述历史申请信息包括：历史申请的专利案件的技术领域、历史合作的代理机构及代理人；在采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息之前，还包括：根据所述申请人信息从所述第三方数据库中抓取与所述申请人信息对应的所有专利案件数据。
其中，将申请人信息作为搜索条件，从第三方数据库中抓取与该申请人对应的所有专利案件数据。从而可以保障抓取到的专利案件数据是准确的。
在一个实施例中，采用预设的专利分析模型对申请人信息对应的所有专利案件数据进行分析，分析得到申请人信息对应的历史申请信息，历史申请信息包括：历史申请的专利案件的技术领域、历史合作的代理机构以及代理人，包括：采用预设的专利分析模型提取出每个专利案件对应的技术分类号、技术关键字、代理机构名称和代理人；根据技术分类号和技术关键字确定专利案件对应的技术领域；提取每个专利案件对应的代理机构，并统计每个代理机构历史代理的第一案件数量，根据第一案件数量将历史合作的代理机构进行排序；提取每个专利案件对应的代理人，统计每个代理人代理的第二案件数量，根据第二案件数量将历史合作的代理人进行排序。
其中，专利分析模型用于提取每个专利案件的技术分类号、技术关键字、代理机构名称和代理人。技术领域用于确定客户的技术所属领域，比如，是属于电子烟领域，还是智能家居领域等。然后代理机构信息为了获取客户偏好合作的代理机构。代理人信息为了获取客户偏好合作的代理人。
在一个实施例中，代理需求信息包括：领域需求、级别需求、专业需求、代理机构需求中的至少一个，代理人信息包括：级别信息、领域信息、专业信 息和所属代理机构信息；
根据代理需求信息、历史评价信息和历史申请信息与数据库中代理人信息进行匹配，生成匹配的代理人列表，包括：基于代理需求信息和代理人信息筛选出与代理需求信息匹配的候选代理人，根据代理需求信息、历史评价信息和历史申请信息计算得到与每个候选代理人的匹配度，根据匹配度生成代理人列表，代理人列表中的代理人信息按照匹配度的高低进行排列。
其中，领域可以根据实际情况进行划分，在一个实施例中，可以将领域划分为四个领域，即软件通信、电子、机械和生化。在另一个实施例中，还可以将领域进一步细分，比如，软件通信还可以细分为：纯软件、软硬结合、通信等。级别需求是指代理人的专业级别需求，预先将代理人划分为了多个专业等级，专业等级反映的是代理人的专业水平。专业需求是指代理人所学的专业，比如，专业分为光电子通信、机电自动化等等。代理机构是指代理人所在的代理机构。代理机构需求可以分为两种，一种是明确需求，即指定代理机构，一种是模糊需求，只需要指定代理机构要满足的条件。
在一个实施例中，获取客户的代理需求信息，包括：获取客户输入的代理需求内容，对代理需求内容进行语义分析，提取出代理需求内容的语义信息；基于语义信息确定客户的代理需求信息，代理需求信息包括：领域需求、专业级别需求、专业需求、时限需求中的至少一个。
其中，代理需求内容是客户根据需要输入的内容，输入的方式可以是语音输入，也可以手动输入，比如，客户语音输入“我需要一个电学领域的代理人，要求15天之内返稿”。通过对输入的内容进行语义分析，提取的客户的代理需求信息为：电学领域、15天时限。
在一个实施例中，对代理需求内容进行语义分析，提取出代理需求内容的语义信息，包括：将代理需求内容作为语义分析模型的输入，采用语义分析模型对代理需求内容进行语义信息的提取，语义信息包括：语义关系和语义内容。
其中，语义信息往往用来表示用户意图的意图信息。在一个具体的实施例中，意图信息以三元组、三元组的组合、意图三元组或意图三元组的组合形式展示。在一个实施例中，语义信息包括三元组或三元组的组合。三元组指的是(x，y，z)形式下的结构数据，用以标识x、y、z以及对应的关系。在本实施例中，三元组由一个句法/语义关系以及两个概念、实体、词或词组组成。意图 三元组为以三元组的形式存储的用户意图，为标识完整意图中的一个小的单元，可以标识为(subject，relation，object)，其中，subject为第一实体，relation表示subject和object之间的关系，object表示第二实体。举个例子，比如，我需要电学代理人。采用三元组表示(我，需要关系，电学代理人)。
语义分析模型的训练往往需要构建大量的数据，由于本方案中语义分析模型的应用场景比较特殊，所以对于语义分析模型的训练数据的构建有其特殊性，基于该应用场景的特殊性，提出了一种快速构建语义分析模型的训练数据集的方法。确定领域需求的候选关键词、专业级别需求的候选关键词、专业需求的候选关键词、时限需求的候选关键词；
根据所述领域需求的候选关键词按照预设的模板自动生成包含有所述领域需求的候选关键词的训练语句，然后将相应的领域需求的候选关键词作为训练语句的语义标注；
根据所述专业级别需求的候选关键词按照预设的模板自动生成包含有所述专业级别需求的候选关键词的训练语句，将相应的候选关键词作为训练语句的语义标注；
根据所述专业需求的候选关键词按照预设的模板自动生成包含有所述专业需求的候选关键词的训练语句，将相应的候选关键词作为训练语句的语义标注；
根据所述时限需求的候选关键词按照预设的模板自动生成包含有所述时限需求的候选关键词的训练语句，将相应的候选关键词作为训练语句的语义标注。
通过上述方法，实现了快速构建训练数据集的目的，有利于提高模型训练的速度，同时大大降低了成本。
在一个实施例中，代理人信息包括：专业级别信息、领域信息、专业信息和时间信息；根据代理需求信息、历史评价信息和历史申请信息与数据库中代理人信息进行匹配，生成匹配的代理人列表，包括：基于代理需求信息从代理人信息中筛选出与代理需求信息匹配的候选代理人；根据代理需求信息、历史评价信息和历史申请信息计算得到与每个候选代理人的匹配度；根据匹配度生成代理人列表，代理人列表中的代理人信息按照匹配度的高低进行排列。
其中，代理需求信息中包含的信息是比较精确的需求信息，所以可以先基于代理需求信息筛选出与代理需求信息匹配的代理人作为候选代理人，举个例 子，当代理需求信息中包含有代理人的专业级别时，那么就可以直接筛选出符合该专业级别的代理人作为候选代理人，这样有利于减少后续进行匹配的计算工作量。在得到候选代理人后，还需要进行进一步的匹配，根据代理需求信息、历史评价信息和历史申请信息进行综合匹配度的计算，得到与每个候选代理人之间的匹配度。按照匹配度的高低进行排列有利于客户优先选择匹配度高的代理人。上述过程中，先基于代理需求信息筛选出候选代理人，然后再计算与每个候选代理人的匹配度，这样有利于减少匹配计算的工作量，大大提高了匹配的效率。
如图4所示，在一个实施例中，根据代理需求信息、历史评价信息和历史申请信息计算得到与每个候选代理人的匹配度，包括：
步骤402，根据代理需求信息、历史评价信息和历史申请信息确定领域需求、级别需求、专业需求和时限需求。
其中，代理需求信息中如果包含的需求信息比较全面，比如，既包含有领域需求、也包含代理机构需求，还包括专业需求，那么根据代理需求信息就可以确定这些信息。如果代理需求信息中没有包含那么多信息，则需要根据历史申请信息和历史评价信息分析出所缺少的信息，假设代理需求信息中没有专业需求，那么可以根据历史申请信息中的技术领域分析出客户的专业需求信息。
级别需求可以为空，即用户可以不限制级别，同样地，专业需求也可以为空，时限需求也可以为空，即用户可以不限制专业和时限。但是领域需求不能为空，不同领域的代理人适合代理的案件类型是不同的，所以为了给客户匹配出适合该案件的领域，领域需求不能为空。
步骤404，根据领域需求和候选代理人对应的领域信息进行相似度计算，得到领域需求对应的领域匹配度。
其中，领域匹配度可以采用领域相似度的方式计算得到，在一个实施例中，采用如下公式计算得到领域匹配度，首先，将领域需求表示成向量的形式，同样地，代理人的领域信息也表示成向量的形式。
Figure PCTCN2022098860-appb-000001
其中，D表示领域相似度，x i表示领域需求向量中的第i个特征值，y i表示领域信息向量中的第i个特征值。
步骤406，根据专业需求和候选代理人对应的专业信息进行相关度计算， 得到专业需求对应的专业匹配度。
其中，专业匹配度可以采用专业相似度的方式计算得到，在一个实施例中，可以采用如下公式计算得到：
Figure PCTCN2022098860-appb-000002
其中，d表示专业相似度，x i表示专业需求向量中的第i个特征值，y i表示专业信息向量中的第i个特征值。
步骤408，根据级别需求和候选代理人对应的级别信息进行匹配，得到级别需求对应的级别匹配度。
其中，级别之间的匹配规则可以预先设置，比如，相同级别的匹配度为100％，级别的匹配可以就高不就低，比如，客户如果要求级别为二级，那么低于二级的匹配度设置为0，而高于二级的匹配度可以逐渐降低，比如，三级与二级的匹配度为80％，四级与二级的匹配度为60％等。
步骤410，根据时限需求和候选代理人对应的时间信息进行匹配，得到时限需求对应的时限匹配度。
其中，时限之间的匹配度的规则可以根据需要设置，匹配的原则可以就低不就高，比如，客户要求时限是10天返稿，那么时间信息可以是小于10天之内的返稿，如果超过10天，相应的匹配度设置为0。
步骤412，根据领域匹配度、专业匹配度、级别匹配度和时限匹配度确定与每个候选代理人的匹配度。
其中，在已知领域匹配度、专业匹配度、级别匹配度和时限匹配度后，就可以综合多个匹配度得到一个综合匹配度，将综合匹配度作为与候选代理人之间的匹配度。综合多个因素进行匹配，有利于为客户提供更加合适的代理人。
在一个实施例中，根据领域匹配度、专业匹配度、级别匹配度和时限匹配度确定与每个候选代理人的匹配度，包括：从代理需求信息和历史评价信息中获取客户的偏重信息，偏重信息包括：客户对领域需求、专业需求、级别需求、时限需求的偏重程度；将偏重信息作为权重分析模型的输入，获取权重分析模型输出的与领域匹配度对应的领域权重、与专业匹配度对应的专业权重，与级别需求对应的级别权重和与时限需求对应的时限权重；根据领域匹配度、领域权重、专业匹配度、专业权重、级别需求、级别权重、时限需求与时限权重计算得到与每个候选代理人的匹配度。
其中，不同的客户对各个因素的偏重是不同的，所以为了匹配出更加合适 的代理人，需要根据客户的偏重信息利用权重分析模型来确定各个因素的权重。继而再利用加权求和的方式计算得到与每个候选代理人的匹配度。
为了迅速地为客户推荐匹配的专利代理师，在信息处理平台预先建立代理人信息库和客户信息库，这样，当客户需要进行代理人推荐时，获取到代理人需求信息后，就可以直接从客户信息库中获取历史申请信息，
如图5所示，在一个实施例中，一种信息交互方法，应用于信息处理平台，包括：
步骤502，建立代理人信息库，代理人信息库中包括代理人专业级别。
其中，代理人信息库中存储了代理人的个人介绍信息，代理人的个人介绍信息包括：代理人专业级别。此外，代理人的个人介绍信息还可以包括：历史代理的主要申请人和历史代理的主要技术领域。代理人的个人介绍信息通过上述的基于数据交互的信息处理方法得到。
步骤504，建立客户信息库，客户信息库中包括：历史申请信息。
其中，客户信息库中存储了很多个客户信息，每个客户信息包括：历史申请信息，历史申请信息是指通过对客户历史申请的专利文件数据进行分析得到的信息，包括：历史申请的专利案件的技术领域、历史合作的代理机构以及代理人中的至少一种。
在一个实施例中，客户信息库中还包括：历史评价信息；历史评价信息是指客户对于历史合作的代理人的评价，包括：专业度的评价和服务态度的评价等。
所述根据所述代理需求信息、所述目标客户对应的历史申请信息与代理人信息库中的代理人信息进行匹配，生成匹配的代理人列表，包括：
根据所述代理需求信息、所述目标客户对应的历史申请信息和历史评价信息与所述代理人信息库中的代理人信息进行匹配，生成匹配的代理人列表。即同时根据代理需求信息、历史申请信息和历史评价信息与代理人信息进行匹配，得到代理人列表。
步骤506，获取目标客户的代理需求信息，根据代理需求信息和目标客户对应的历史申请信息与代理人信息库中的代理人信息进行匹配，生成匹配的代理人列表，代理人列表中包含多个匹配的代理人信息。
其中，代理需求信息是指客户输入的较为精确的需求信息，代理需求信息包括：对代理人的专业级别要求、对代理人的专业要求、对代理人的执业年限的要求、对于案件的完成时限的要求等中的一个或多个。综合代理需求信息、和历史申请信息，然后与数据库中的代理人信息进行匹配，生成代理人列表，在代理人列表中可以按照匹配度从高到低进行排序，从而便于后续客户进行选择。
在另一个实施例中，还包括：获取目标客户的历史评价信息，根据代理需求信息和目标客户对应的历史申请信息和历史评价信息与代理人信息库中的代理人信息进行匹配，生成匹配的代理人列表。
步骤508，将生成的代理人列表返回给客户终端。
其中，将代理人列表返回给客户终端，客户根据代理人列表进行选择，大大提高了选择的准确度。
上述信息交互方法，在信息处理平台预先建立代理人信息库和客户信息库，这样，当客户需要进行代理人推荐时，获取到代理人需求信息后，就可以直接从客户信息库中获取历史申请信息和历史评价信息，然后基于代理需求信息、历史申请信息和历史评价信息等三个方面的信息与代理人信息库中的代理人信息进行匹配，得到匹配的代理人列表，然后将代理人列表返回给客户终端作为推荐结果，该信息交互方法中由于预先建立了代理人信息库和客户信息库，实现了迅速且准确地为客户推荐匹配的专利代理师。
在一个实施例中，上述信息交互方法还包括：接收客户终端发送的从代理人列表中选定的意向代理人，将客户的意向请求发送给意向代理人终端；接收意向代理人终端发送的意向结果；当意向结果为同意时，建立客户终端与代理人终端之间的即时通信连接通道，即时通信连接通道用于客户与代理人之间进行案件沟通；将客户与代理人之前的案件沟通过程保存为案件沟通过程文件，案件沟通过程文件与对应的专利案件关联存储。
其中，当将代理人列表推送到客户终端后，客户从代理人列表中选定意向代理人。由于在该信息处理平台上是双向选择的，所以还需要将意向请求发送给意向代理人终端去确认，只有意向代理人同意委托后，才建立客户终端与代理人终端之间的即时通信连接通道，客户与代理人可以通过该即时通信连接通道进行案件沟通交流，并保存沟通过程文件，便于后续出现分歧时进行复盘， 同时也便于代理人反复查看沟通过程文件来理解方案。通过将案件沟通过程文件与对应的专利进行关联存储，也便于后续查看与该案件对应的沟通记录，而不需要从众多的信息中去查找。
目前所有的即时通信工具的信息都是统一保存的，当需要查看某个时间点或某个事件时，需要从众多历史聊天记录中查找，效率低下。而该方法中创新性地将案件沟通过程文件与对应的专利案件关联存储，使得后续可以迅速查看每个案件的沟通过程。
在一个实施例中，上述上述信息交互方法还包括：将案件沟通过程文件进行加密保存，以保证客户信息的安全性。
在一个实施例中，上述方法还包括：建立客户案件信息库，包括：通过与第三方数据库进行交互，从第三方数据库中抓取与客户信息对应的已公开的专利案件信息，已公开的专利案件信息包括：案件审查状态；获取客户信息对应的待处理的专利案件信息，待处理的专利案件信息包括：案件处理状态；将同一客户对应的已公开的专利案件信息和待处理的案件信息存储到同一客户对应的客户案件信息库，一个客户对应一个客户案件信息库。
其中，为了方便客户对自己的管件管理查看，在信息处理平台，为每个客户建立一个客户案件信息库，将该客户已公开的专利案件数据从第三方数据库进行抓取并存储该客户案件信息库中，此外，还将客户对应的待处理的专利案件信息加入到客户案件信息库，使得客户通过登录该信息处理平台，便可快速查看所有案件的状态。这样，客户通过信息处理平台便可以直接对案件进行管理，使得客户可以更加迅速且准确的了解每个案件的状态。相对于以往如果想要了解案件状态，只能向代理机构进行问询才能查到的方式，该方式无疑更加快捷和准确。
在一个实施例中，上述方法还包括：接收客户终端发送的查看案件请求，响应于请求获取客户对应的权限信息，从客户案件信息库中获取与权限信息对应的专利案件信息返回到客户终端进行展示。
其中，客户终端登录信息处理平台后，发送查看案件请求，触发查看案件请求的方式有很多，比如，可以直接点击案件查询的按钮。
不同的客户对应的权限是不一样的，有些客户的权限可以查看所有案件信息，而有些客户的权限只能查看他自己对应的案件。举个例子，对于同一个公 司而言，分为经理和普通员工，经理的权限可以是查看所有普通员工的案件，而普通员工只能查看自己对应的案件。权限的设置有利于信息的保密。上述过程中，在获取到案件请求后，首先确定客户对应的权限，然后与该权限信息对应的专利案件信息返回到客户终端进行展示。该方式大大提高了案件查询的灵活性，且在查询的过程中通过设置权限信息有利于信息安全。
在一个实施例中，已公开的专利案件信息还包括：专利审查的过程文件，过程文件包括受理通知书、补正通知书、审查意见通知书、审查答复文件中的一种或多种；待处理的专利案件信息包括：专利处理的过程文件，专利处理的过程文件包括技术交底书、案件沟通过程文件、案件撰写稿件中的一种或多种。
其中，上述提到的已公开的专利案件信息还包括：专利审查过程中产生的过程文件。过程文件包括：受理通知书、补正通知书、审查意见通知书、审查答复文件中的一种或多种。具体地，信息处理平台可以直接接收官方下发的审查过程文件(受理通知书、补正通知书、审查意见通知书)，然后将接收到的审查过程文件与对应的案件进行关联存储。这样客户在该信息处理平台便可以查询得到所有相关的文件，方便快捷。
在一个实施例中，建立代理人信息库，包括：获取代理人的基础信息，基础信息包括：代理人姓名和执业经历，执业经历中包括代理人在执业期间对应的代理机构名称；与第三方数据库建立连接，第三方数据库中存储有专利案件信息，专利案件信息中包括：代理机构名称和代理人姓名；根据代理人姓名和执业经历从第三方数据库中提取出与代理人对应的所有专利案件数据；采用建立的专利案件数据处理模型对代理人对应的所有专利案件数据进行处理，统计出代理人的历史授权率；从代理人对应的所有专利案件数据中抽取出预设年限内的预设数量的待评价专利案件；对待评价专利案件进行技术分析，确定待评价专利案件所属的技术领域；根据技术领域将待评价专利案件发送给技术领域对应的专家进行评审，接收返回的评审结果；获取代理人对应的历史服务评价信息；根据代理人的历史授权率、评审结果和历史服务评价信息按照预设算法确定代理人的专业级别；对代理人对应的所有专利案件数据进行分析处理，统计得到代理人历史代理的主要申请人和历史代理的主要技术领域；根据代理人的专业级别、历史代理的主要申请人和历史代理的主要技术领域按照预设模板生成代理人对应的个人介绍信息；将代理人对应的个人介绍信息存入代理人信 息库中。
在一个实施例中，建立客户信息库，包括：获取客户的基础信息，基础信息包括：申请人信息；与第三方数据库建立连接，第三方数据库中存储有专利案件信息；根据申请人信息从第三方数据库中抓取与申请人信息对应的所有专利案件数据；采用预设的专利分析模型对申请人信息对应的所有专利案件数据进行分析，分析得到申请人信息对应的历史申请信息，历史申请信息包括：历史申请的专利案件的技术领域、历史合作的代理机构以及代理人；将客户对应的历史申请信息存入客户信息库中。
如图6所示，为一个信息交互的时序图。包括两个部分，第一部分，生成代理人的个人介绍信息的时序图，第二部分是进行代理人推荐的时序图。
第一部分：首先，代理人通过代理人终端进行注册，注册过程中需要填写代理人基础信息，代理人基础信息包括：代理人姓名和执业经历。信息处理平台接收到代理人基础信息后，向第三方数据库发送数据请求(携带代理人姓名+执业经历)，并接收第三方数据库返回的专利案件数据。然后信息处理平台对专利案件数据进行分析，统计得到历史授权率，此外，信息处理平台还从专利案件数据中抽取出预设年限内的预设数量的待评价专利案件，并进行技术分析，确定待评价专利案件所属的技术领域；之后，将待评价专利案件发送给对应的专家进行评审，接收返回的评审结果；此外，信息处理平台从自身数据库中获取代理人对应的历史服务评价信息；继而根据代理人的历史授权率、评审结果和历史服务评价信息确定代理人的专业级别；另外，信息处理平台通过对代理人对应的所有专利案件数据进行分析处理，统计得到所述代理人历史代理的申请人和历史代理的技术领域；最后，根据所述代理人的专业级别、所述历史代理的主要申请人和历史代理的主要技术领域按照预设模板生成所述代理人对应的个人介绍信息，并将生成的个人介绍信息存入代理人信息库中。
第二部分：客户通过客户终端在信息处理平台进行注册，注册时填写客户的基础信息(至少包括申请人信息)，信息处理平台根据申请人信息从第三方数据库中抓取相应的专利案件数据，并采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息，所述历史申请信息包括：历史申请的专利案件的技术领域、历史合作 的代理机构以及代理人，加入到客户信息库中；此外，信息处理平台还获取客户终端发送的代理需求信息，并从自身数据库中获取客户对应的历史评价信息，最后根据代理需求信息、历史评价信息和历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息；将生成的所述代理人列表返回给客户终端。客户终端选择意向代理人，然后信息处理平台发送给代理人终端进行确认，代理人同意后达成一致，
如图7所示，提出了一种基于数据交互的信息处理装置，该装置包括：
获取模块702，用于获取代理人的基础信息，所述基础信息包括：代理人身份识别信息；
连接模块704，用于与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息，所述专利案件信息中包括：代理人身份识别信息；
处理模块706，用于采用建立的专利案件数据处理模型对所述代理人对应的所有专利案件数据进行处理，统计出所述代理人在预设时间内的历史授权率；
抽取模块708，用于从所述代理人对应的所有专利案件数据中抽取出预设年限内的预设数量的待评价专利案件；
评审模块710，用于将所述待评价专利案件发送给所述技术领域对应的专家进行评审，接收返回的评审结果；
确定模块712，用于根据所述代理人的历史授权率和所述评审结果按照预设算法确定所述代理人的专业级别；
生成模块714，用于根据所述代理人的专业级别按照预设模板生成所述代理人对应的个人介绍信息。
在一个实施例中，上述装置还包括：提取模块，用于根据所述代理人姓名和所述执业经历从所述第三方数据库中提取出所述代理人对应的所有专利案件数据。
在一个实施例中，所述基础信息包括：专业代理方向；抽取模块708还用于从代理人对应的所述所有专利案件数据中筛选出与所述专业代理方向匹配的候选专利案件数据；从所述候选专利案件数据中抽取出待评价专利案件。
在一个实施例中，处理模块706还用于采用建立的所述专利案件数据处理模型提取所述所有专利案件数据中法律状态，所述法律状态包括：授权、审中、 驳回、撤回中的一种，所述专利案件数据处理模型用于确定法律状态对应的正则表达式，基于所述正则表达式进行各个案件的法律状态的提取；统计法律状态处于授权状态的授权案件数量、所述法律状态处于驳回状态的驳回案件数量和所述法律状态处于撤回状态的撤回案件数量；根据所述授权案件数量、所述驳回案件数量和所述撤回案件数量计算得到所述历史授权率。
在一个实施例中，处理模块706还用于统计距离当前时间最近的预设数量的已结案案件中的授权案件数量、驳回案件数量和撤回案件数量；根据所述距离当前时间最近的预设数量的已结案案件中的授权案件数量所述驳回案件数量和撤回案件数量计算得到最新的所述历史授权率。
在一个实施例中，处理模块706还用于筛选出申请日在预设时间段内的候选专利案件；统计所述候选专利案件中处于授权状态的授权案件数量、所述处于驳回状态的驳回案件数量和所述处于撤回状态的撤回案件数量；根据所述候选专利案件中处于授权状态的授权案件数量、所述处于驳回状态的驳回案件数量和所述处于撤回状态的撤回案件数量计算得到所述预设时间段对应的历史授权率。
在一个实施例中，确定模块712还用于获取所述代理人的从业年限和总的历史案件申请量，将所述从业年限和所述总的历史案件申请量作为权重确定模型的输入，所述权重确定模型是采用深度神经网络模型进行训练学习得到的；获取所述权重确定模型输出的所述代理人对应的所述历史授权率的第一权重，所述评审结果的第二权重和所述历史服务评价信息的第三权重；根据所述历史授权率和所述第一权重，所述评审结果和所述第二权重、所述历史服务评价信息和所述第三权重确定所述代理人的专业级别。
在一个实施例中，提取模块还用于根据所述代理人姓名和所述执业经历确定目标搜索条件，所述目标搜索条件中包括代理人姓名、执业的代理机构名称以及对应的执业时间；根据所述目标搜索条件从所述第三方数据库中提取出所述代理人对应的所有专利案件数据。
在一个实施例中，上述装置还包括：分析模块，用于对所述代理人对应的所有专利案件数据进行分析处理，统计得到所述代理人历史代理的申请人和历史代理的技术领域。
所述分析模块还用于抓取每件专利案件的分类号信息和摘要信息，对所述 摘要信息进行分析，提取出技术关键词；将所述分类号信息和所述技术关键词作为技术领域分类模型的输入，确定所述专利案件对应的技术领域类别，将所述确定的技术领域类别作为所述专利案件的技术领域标签。
在一个实施例中，所述技术领域分类模型包括：第一特征模型、第二特征模型和分类模型，第一特征模型用于根据分类号确定所述分类号对应的第一特征向量，第二特征模型用于确定所述技术关键词对应的第二特征向量，所述分类模型用于根据所述第一特征向量和第二特征向量确定所述专利案件对应的技术领域类别。
在一个实施例中，所述基础信息还包括：代理人资格证号，上述装置还包括：
校验模块，用于根据所述代理人资格证号获取所述代理人的官方执业经历；根据所述官方执业经历对所述代理人的基础信息进行校验，若校验成功，则通知连接模块进入与第三方数据库建立连接；若校验失败，则返回校验失败的结果。
在一个实施例中，统计模块还用于提取每一件案子中的申请人信息，统计同一申请人信息对应的案件数量；根据每个申请人信息对应的案件数量确定主要申请人，所述主要申请人为案件数量排名靠前的申请人；获取每一件案子的技术分类号和技术关键字，根据所述技术分类号和所述技术关键字确定每一件案子对应的技术领域；统计同一技术领域对应的案件数量，根据所述同一技术领域对应的案件数量确定所述主要技术领域。
如图8所示，提出了一种基于数据交互的信息推荐装置，包括：
获取模块802，用于获取客户的基础信息，所述客户的基础信息包括：申请人信息；
连接模块804，用于与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息；
分析模块806，用于采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息，所述历史申请信息包括：历史申请的专利案件的技术领域、历史合作的代理机构以及代理人；
所述获取模块还用于获取客户的代理需求信息；
匹配模块808，用于根据所述代理需求信息和所述历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息；
返回模块810，用于将生成的所述代理人列表返回给客户终端。
在一个实施例中，上述装置还包括：抓取模块，用于根据所述申请人信息从所述第三方数据库中抓取与所述申请人信息对应的所有专利案件数据。
在一个实施例中，分析模块还用于采用所述预设的专利分析模型提取出每个专利案件对应的技术分类号、技术关键字、代理机构名称和代理人；根据所述技术分类号和技术关键字确定所述专利案件对应的技术领域；提取每个专利案件对应的代理机构，并统计每个代理机构历史代理的第一案件数量，根据所述第一案件数量将历史合作的所述代理机构进行排序；提取每个专利案件对应的代理人，统计每个代理人代理的第二案件数量，根据所述第二案件数量将历史合作的代理人进行排序。
在一个实施例中，获取模块还用于获取客户输入的代理需求内容，对所述代理需求内容进行语义分析，提取出所述代理需求内容的语义信息；基于所述语义信息确定所述客户的代理需求信息，所述代理需求信息包括：领域需求、专业级别需求、专业需求、时限需求中的至少一个。
在一个实施例中，所述代理人信息包括：专业级别信息、领域信息、专业信息和时间信息；匹配模块还用于基于所述代理需求信息从所述代理人信息中筛选出与所述代理需求信息匹配的候选代理人；根据所述代理需求信息、所述历史评价信息和所述历史申请信息计算得到与每个所述候选代理人的匹配度；根据匹配度生成代理人列表，所述代理人列表中的代理人信息按照匹配度的高低进行排列。
在一个实施例中，匹配模块还用于根据所述代理需求信息、所述历史评价信息和所述历史申请信息确定领域需求、级别需求、专业需求和时限需求；根据所述领域需求和所述候选代理人对应的领域信息进行相似度计算，得到所述领域需求对应的领域匹配度；根据所述专业需求和所述候选代理人对应的专业信息进行相关度计算，得到所述专业需求对应的专业匹配度；根据所述级别需求和所述候选代理人对应的级别信息进行匹配，得到所述级别需求对应的级别 匹配度；根据所述时限需求和所述候选代理人对应的时间信息进行匹配，得到所述时限需求对应的时限匹配度；根据所述领域匹配度、所述专业匹配度、级别匹配度和所述时限匹配度确定与每个所述候选代理人的匹配度。
在一个实施例中，匹配模块还用于从所述代理需求信息和所述历史评价信息中获取客户的偏重信息，所述偏重信息包括：客户对所述领域需求、所述专业需求、所述级别需求、所述时限需求的偏重程度；将所述偏重信息作为权重分析模型的输入，获取所述权重分析模型输出的与所述领域匹配度对应的领域权重、与所述专业匹配度对应的专业权重，与所述级别需求对应的级别权重和与所述时限需求对应的时限权重；根据所述领域匹配度、领域权重、所述专业匹配度、专业权重、所述级别需求、所述级别权重、时限需求与所述时限权重计算得到与每个所述候选代理人的匹配度。
在一个实施例中，所述代理人信息包括：级别信息；所述级别信息是通过以下方式确定的：获取代理人的基础信息，所述代理人的基础信息包括：代理人姓名和执业经历，所述执业经历中包括代理人在执业期间对应的代理机构名称；与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息，所述专利案件信息中包括：代理机构名称和代理人姓名；根据所述代理人姓名和所述执业经历从所述第三方数据库中提取出与所述代理人对应的所有专利案件数据；采用建立的专利案件数据处理模型对所述代理人对应的所有专利案件数据进行处理，统计出所述代理人的历史授权率；从所述代理人对应的所有专利案件数据中抽取出预设年限内的预设数量的待评价专利案件；对所述待评价专利案件进行技术分析，确定所述待评价专利案件所属的技术领域；根据所述技术领域将所述待评价专利案件发送给所述技术领域对应的专家进行评审，接收返回的评审结果；获取所述代理人对应的历史服务评价信息；根据所述代理人的历史授权率、所述评审结果和所述历史服务评价信息按照预设算法确定所述代理人的专业级别。
在一个实施例中，上述基于数据交互的信息推荐还包括：通信建立模块，用于接收所述客户终端发送的从所述代理人列表中选定的意向代理人，将客户的意向请求发送给所述意向代理人终端；接收所述意向代理人终端发送的意向结果，当所述意向结果为同意时，建立所述客户终端与所述代理人终端之间的即时通信连接通道，所述即时通信连接通道用于客户与代理人之间进行案件沟 通。
如图9所示，提出了一种信息交互装置，包括：
第一建立模块902，用于建立代理人信息库，所述代理人信息库中包括多个代理人信息，每个所述代理人信息包括：代理人专业级别；
第二建立模块904，用于建立客户信息库，所述客户信息库中包括：多个客户信息，每个所述客户信息包括：历史申请信息；
获取模块906，用于获取目标客户的代理需求信息；
匹配模块908，用于根据所述代理需求信息、所述目标客户对应的历史申请信息与代理人信息库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息；
返回模块910，用于将生成的所述代理人列表返回给客户终端。
在一个实施例中，上述信息交互装置还包括：
接收模块，用于接收所述客户终端发送的从所述代理人列表中选定的意向代理人，将客户的意向请求发送给所述意向代理人终端；接收所述意向代理人终端发送的意向结果；
通信建立模块，用于当所述意向结果为同意时，建立所述客户终端与所述代理人终端之间的即时通信连接通道，所述即时通信连接通道用于客户与代理人之间进行案件沟通；
保存模块，用于将所述客户与代理人之前的案件沟通过程保存为所述案件沟通过程文件，所述案件沟通过程文件与对应的专利案件关联存储。
在一个实施例中，第一建立模块还用于通过与第三方数据库进行交互，从第三方数据库中抓取与所述客户信息对应的已公开的专利案件信息，所述已公开的专利案件信息包括：案件审查状态；获取所述客户信息对应的待处理的专利案件信息，所述待处理的专利案件信息包括：案件处理状态；将同一客户对应的所述已公开的专利案件信息和所述待处理的案件信息存储到所述同一客户对应的客户案件信息库，一个客户对应一个客户案件信息库。
在一个实施例中，上述信息交互装置还包括：
查找模块用于接收客户终端发送的查看案件请求，响应于所述请求获取所述客户对应的权限信息，从所述客户案件信息库中获取与所述权限信息对应的 专利案件信息返回到所述客户终端进行展示。
在一个实施例中，所述已公开的专利案件信息还包括：专利审查的过程文件，所述过程文件包括受理通知书、补正通知书、审查意见通知书、审查答复文件中的一种或多种；所述待处理的专利案件信息包括：专利处理的过程文件，所述专利处理的过程文件包括技术交底书、案件沟通过程文件、案件撰写稿件中的一种或多种。
在一个实施例中，第二建立模块还用于获取代理人的基础信息，所述基础信息包括：代理人姓名和执业经历，所述执业经历中包括代理人在执业期间对应的代理机构名称；与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息，所述专利案件信息中包括：代理机构名称和代理人姓名；根据所述代理人姓名和所述执业经历从所述第三方数据库中提取出与所述代理人对应的所有专利案件数据；采用建立的专利案件数据处理模型对所述代理人对应的所有专利案件数据进行处理，统计出所述代理人的历史授权率；从所述代理人对应的所有专利案件数据中抽取出预设年限内的预设数量的待评价专利案件；对所述待评价专利案件进行技术分析，确定所述待评价专利案件所属的技术领域；根据所述技术领域将所述待评价专利案件发送给所述技术领域对应的专家进行评审，接收返回的评审结果；获取所述代理人对应的历史服务评价信息；根据所述代理人的历史授权率、所述评审结果和所述历史服务评价信息按照预设算法确定所述代理人的专业级别；对所述代理人对应的所有专利案件数据进行分析处理，统计得到所述代理人历史代理的主要申请人和历史代理的主要技术领域；根据所述代理人的专业级别、所述历史代理的主要申请人和历史代理的主要技术领域按照预设模板生成所述代理人对应的个人介绍信息；将所述代理人对应的个人介绍信息存入所述代理人信息库中。
在一个实施例中，上述装置还包括：第三建立模块，用于获取客户的基础信息，所述基础信息包括：申请人信息；与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息；根据所述申请人信息从所述第三方数据库中抓取与所述申请人信息对应的所有专利案件数据；采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息，所述历史申请信息包括：历史申请的专利案件的技术领域、历史合作的代理机构以及代理人；将所述客户对应的历史申请信息存入所 述客户信息库中。
图10示出了一个实施例中计算机设备的内部结构图。该计算机设备可以为服务器，该计算机设备包括通过系统总线连接的处理器和存储器。其中，存储器包括非易失性存储介质和内存储器。该计算机设备的非易失性存储介质存储有操作系统，还可存储有计算机程序，该计算机程序被处理器执行时，可使得处理器实现上述基于数据交互的信息处理方法或基于数据交互的信息推荐方法或信息交互方法。该内存储器中也可储存有计算机程序，该计算机程序被处理器执行时，可使得处理器执行上述基于数据交互的信息处理方法或基于数据交互的信息推荐方法或信息交互方法。本领域技术人员可以理解，图10中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备的限定，具体的计算机设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
一种计算机可读存储介质，存储有计算机程序，所述计算机程序被处理器执行时，使得所述处理器执行上述基于数据交互的信息处理方法或基于数据交互的信息推荐方法或信息交互方法的步骤。
一种计算机设备，包括存储器和处理器，所述存储器有存储计算机程序，所述计算机程序被所述处理器执行时，使得所述处理器执行上述基于数据交互的信息处理方法或基于数据交互的信息推荐方法或信息交互方法的步骤。
需要说明的是，上述基于数据交互的信息处理方法、装置、计算机设备及存储介质，基于数据交互的信息推荐方法、装置、计算机设备及存储介质以及信息交互方法、装置、计算机设备及存储介质具有相同或相应的技术特征，上述相应的实施例可以相互适用。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的程序可存储于一非易失性计算机可读取存储介质中，该程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和/或易失性存储器。非易失性存 储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限，RAM以多种形式可得，诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对本申请专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。

Claims (13)

  1. 一种基于数据交互的信息推荐方法，应用于信息处理平台，其特征在于，包括：
    获取客户的基础信息，所述客户的基础信息包括：申请人信息；
    与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息；
    采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息；
    获取客户的代理需求信息；
    根据所述代理需求信息和所述历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息；
    将生成的所述代理人列表返回给客户终端。
  2. 根据权利要求1所述的方法，其特征在于，所述历史申请信息包括：历史申请的专利案件的技术领域、历史合作的代理机构及代理人；
    在采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息之前，还包括：
    根据所述申请人信息从所述第三方数据库中抓取与所述申请人信息对应的所有专利案件数据。
  3. 根据权利要求1所述的方法，其特征在于，所述方法还包括：获取客户的历史评价信息；
    所述根据所述代理需求信息和所述历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息，包括：
    根据所述代理需求信息、所述历史评价信息和所述历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息。
  4. 根据权利要求1所述的方法，其特征在于，所述历史申请信息包括：历史申请的专利案件的技术领域、历史合作的代理机构以及代理人，所述采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息，包括：
    采用所述预设的专利分析模型提取出每个专利案件对应的技术分类号、技 术关键字、代理机构名称和代理人；
    根据所述技术分类号和技术关键字确定所述专利案件对应的技术领域；
    提取每个专利案件对应的代理机构，并统计每个代理机构历史代理的第一案件数量，根据所述第一案件数量将历史合作的所述代理机构进行排序；
    提取每个专利案件对应的代理人，统计每个代理人代理的第二案件数量，根据所述第二案件数量将历史合作的代理人进行排序。
  5. 根据权利要求1所述的方法，其特征在于，所述获取客户的代理需求信息，包括：
    获取客户输入的代理需求内容，对所述代理需求内容进行语义分析，提取出所述代理需求内容的语义信息；
    基于所述语义信息确定所述客户的代理需求信息，所述代理需求信息包括：领域需求、专业级别需求、专业需求、时限需求中的至少一个。
  6. 根据权利要求3所述的方法，其特征在于，所述代理人信息包括：专业级别信息、领域信息、专业信息和时间信息；
    所述根据所述代理需求信息、历史评价信息和所述历史申请信息与数据库中代理人信息进行匹配，生成匹配的代理人列表，包括：
    基于所述代理需求信息从所述代理人信息中筛选出与所述代理需求信息匹配的候选代理人；
    根据所述代理需求信息、所述历史评价信息和所述历史申请信息计算得到与每个所述候选代理人的匹配度；
    根据匹配度生成代理人列表，所述代理人列表中的代理人信息按照匹配度的高低进行排列。
  7. 根据权利要求6所述的方法，其特征在于，所述根据所述代理需求信息、所述历史评价信息和所述历史申请信息计算得到与每个所述候选代理人的匹配度，包括：
    根据所述代理需求信息、所述历史评价信息和所述历史申请信息确定领域需求、级别需求、专业需求和时限需求；
    根据所述领域需求和所述候选代理人对应的领域信息进行相似度计算，得到所述领域需求对应的领域匹配度；
    根据所述专业需求和所述候选代理人对应的专业信息进行相关度计算，得 到所述专业需求对应的专业匹配度；
    根据所述级别需求和所述候选代理人对应的级别信息进行匹配，得到所述级别需求对应的级别匹配度；
    根据所述时限需求和所述候选代理人对应的时间信息进行匹配，得到所述时限需求对应的时限匹配度；
    根据所述领域匹配度、所述专业匹配度、级别匹配度和所述时限匹配度确定与每个所述候选代理人的匹配度。
  8. 根据权利要求7所述的方法，其特征在于，所述根据所述领域匹配度、所述专业匹配度、级别匹配度和所述时限匹配度确定与每个所述候选代理人的匹配度，包括：
    从所述代理需求信息和所述历史评价信息中获取客户的偏重信息，所述偏重信息包括：客户对所述领域需求、所述专业需求、所述级别需求、所述时限需求的偏重程度；
    将所述偏重信息作为权重分析模型的输入，获取所述权重分析模型输出的与所述领域匹配度对应的领域权重、与所述专业匹配度对应的专业权重，与所述级别需求对应的级别权重和与所述时限需求对应的时限权重；
    根据所述领域匹配度、领域权重、所述专业匹配度、专业权重、所述级别需求、所述级别权重、时限需求与所述时限权重计算得到与每个所述候选代理人的匹配度。
  9. 根据权利要求1所述的方法，其特征在于，所述代理人信息包括：级别信息；
    所述级别信息是通过以下方式确定的：
    获取代理人的基础信息，所述代理人的基础信息包括：代理人姓名和执业经历，所述执业经历中包括代理人在执业期间对应的代理机构名称；
    与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息，所述专利案件信息中包括：代理机构名称和代理人姓名；
    根据所述代理人姓名和所述执业经历从所述第三方数据库中提取出与所述代理人对应的所有专利案件数据；
    采用建立的专利案件数据处理模型对所述代理人对应的所有专利案件数据进行处理，统计出所述代理人的历史授权率；
    从所述代理人对应的所有专利案件数据中抽取出预设年限内的预设数量的待评价专利案件；
    对所述待评价专利案件进行技术分析，确定所述待评价专利案件所属的技术领域；
    根据所述技术领域将所述待评价专利案件发送给所述技术领域对应的专家进行评审，接收返回的评审结果；
    获取所述代理人对应的历史服务评价信息；
    根据所述代理人的历史授权率、所述评审结果和所述历史服务评价信息按照预设算法确定所述代理人的专业级别。
  10. 根据权利要求1所述的方法，其特征在于，所述方法还包括：
    接收所述客户终端发送的从所述代理人列表中选定的意向代理人，将客户的意向请求发送给所述意向代理人终端；
    接收所述意向代理人终端发送的意向结果，当所述意向结果为同意时，建立所述客户终端与所述代理人终端之间的即时通信连接通道，所述即时通信连接通道用于客户与代理人之间进行案件沟通。
  11. 一种基于数据交互的信息推荐装置，其特征在于，包括：
    获取模块，用于获取客户的基础信息，所述客户的基础信息包括：申请人信息；
    连接模块，用于与第三方数据库建立连接，所述第三方数据库中存储有专利案件信息；
    分析模块，用于采用预设的专利分析模型对所述申请人信息对应的所有专利案件数据进行分析，分析得到所述申请人信息对应的历史申请信息；
    所述获取模块还用于获取客户的代理需求信息；
    匹配模块，用于根据所述代理需求信息和所述历史申请信息与数据库中的代理人信息进行匹配，生成匹配的代理人列表，所述代理人列表中包含多个匹配的代理人信息；
    返回模块，用于将生成的所述代理人列表返回给客户终端。
  12. 一种计算机可读存储介质，存储有计算机程序，所述计算机程序被处 理器执行时，使得所述处理器执行如权利要求1至10中任一项所述的基于数据交互的信息推荐方法的步骤。
  13. 一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述计算机程序被所述处理器执行时，使得所述处理器执行如权利要求1至10中任一项所述的基于数据交互的信息推荐方法的步骤。
PCT/CN2022/098860 2021-06-18 2022-06-15 基于数据交互的信息推荐方法、装置、设备及存储介质 WO2022262752A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP22824229.3A EP4358005A1 (en) 2021-06-18 2022-06-15 Information recommendation method and apparatus based on data interaction, and device and storage medium

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
CN202110683059 2021-06-18
CN202110683059.5 2021-06-18
CN202110859616.4A CN113688311A (zh) 2021-06-18 2021-07-28 基于数据交互的信息推荐方法、装置、设备及存储介质
CN202110859616.4 2021-07-28

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US18/544,324 Continuation US20240185370A1 (en) 2021-06-18 2023-12-18 Method for information recommendation based on data interaction, device, and storage medium

Publications (1)

Publication Number Publication Date
WO2022262752A1 true WO2022262752A1 (zh) 2022-12-22

Family

ID=78578129

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/098860 WO2022262752A1 (zh) 2021-06-18 2022-06-15 基于数据交互的信息推荐方法、装置、设备及存储介质

Country Status (3)

Country Link
EP (1) EP4358005A1 (zh)
CN (1) CN113688311A (zh)
WO (1) WO2022262752A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117234095A (zh) * 2023-08-18 2023-12-15 浙江雨林电子科技有限公司 一种全屋家居无线智能控制方法及系统

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113688311A (zh) * 2021-06-18 2021-11-23 诺正集团股份有限公司 基于数据交互的信息推荐方法、装置、设备及存储介质
CN114880323B (zh) * 2022-04-26 2023-12-12 深圳市未来鼠信息技术有限公司 数据管理方法、装置、设备及存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104537552A (zh) * 2014-12-23 2015-04-22 百度在线网络技术（北京）有限公司 通过计算机实现的信息推荐方法及装置
CN107194790A (zh) * 2017-06-21 2017-09-22 苏州发飚智能科技有限公司 个人智能招标数据处理方法及系统
CN109389258A (zh) * 2017-08-02 2019-02-26 北京恒冠网络数据处理有限公司 一种专利申请服务系统及方法
CN110727875A (zh) * 2019-12-17 2020-01-24 杭州实在智能科技有限公司 一种法律案件代理的智能分发方法与系统
US20200327479A1 (en) * 2019-02-20 2020-10-15 David & Raymond Patent Firm Bidirectional Selecting System and Bidirectional Selecting Method between Service Objects and Service Providers
CN113688311A (zh) * 2021-06-18 2021-11-23 诺正集团股份有限公司 基于数据交互的信息推荐方法、装置、设备及存储介质

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105512974A (zh) * 2015-12-01 2016-04-20 北京环球惠通咨询服务有限公司 一种商标信息管理系统及方法
WO2019147804A1 (en) * 2018-01-26 2019-08-01 Ge Inspection Technologies, Lp Generating natural language recommendations based on an industrial language model

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104537552A (zh) * 2014-12-23 2015-04-22 百度在线网络技术（北京）有限公司 通过计算机实现的信息推荐方法及装置
CN107194790A (zh) * 2017-06-21 2017-09-22 苏州发飚智能科技有限公司 个人智能招标数据处理方法及系统
CN109389258A (zh) * 2017-08-02 2019-02-26 北京恒冠网络数据处理有限公司 一种专利申请服务系统及方法
US20200327479A1 (en) * 2019-02-20 2020-10-15 David & Raymond Patent Firm Bidirectional Selecting System and Bidirectional Selecting Method between Service Objects and Service Providers
CN110727875A (zh) * 2019-12-17 2020-01-24 杭州实在智能科技有限公司 一种法律案件代理的智能分发方法与系统
CN113688311A (zh) * 2021-06-18 2021-11-23 诺正集团股份有限公司 基于数据交互的信息推荐方法、装置、设备及存储介质

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117234095A (zh) * 2023-08-18 2023-12-15 浙江雨林电子科技有限公司 一种全屋家居无线智能控制方法及系统
CN117234095B (zh) * 2023-08-18 2024-04-02 浙江雨林电子科技有限公司 一种全屋家居无线智能控制方法及系统

Also Published As

Publication number Publication date
EP4358005A1 (en) 2024-04-24
CN113688311A (zh) 2021-11-23

Similar Documents

Publication Publication Date Title
WO2022262752A1 (zh) 基于数据交互的信息推荐方法、装置、设备及存储介质
WO2020249125A1 (zh) 用于自动训练机器学习模型的方法和系统
CN107851097B (zh) 数据分析系统、数据分析方法、数据分析程序及存储介质
CN112632405B (zh) 一种推荐方法、装置、设备及存储介质
TW201539216A (zh) 文件分析系統、文件分析方法、及文件分析程式
CN111680153A (zh) 一种基于知识图谱的大数据鉴真方法与系统
US20240144405A1 (en) Method for information interaction, device, and storage medium
US20240127143A1 (en) Method, device and storage medium for information processing based on data interaction
CN116109373A (zh) 金融产品的推荐方法、装置、电子设备和介质
CN105786810B (zh) 类目映射关系的建立方法与装置
WO2016016973A1 (ja) 実績評価装置、実績評価装置の制御方法、および実績評価装置の制御プログラム
TW201543237A (zh) 文件分析系統、文件分析方法、以及文件分析程式
WO2016009553A1 (ja) 知財評価システム、知財評価システムの制御方法、および知財評価プログラム
CN115982429B (zh) 一种基于流程控制的知识管理方法及系统
CN116629716A (zh) 一种智能交互系统工效分析方法
CN113610499B (zh) 一种基于区块链的求职功能团队职业信用档案方法和系统
TW201539217A (zh) 文件分析系統、文件分析方法、以及文件分析程式
EP3493082A1 (en) A method of exploring databases of time-stamped data in order to discover dependencies between the data and predict future trends
US20240185370A1 (en) Method for information recommendation based on data interaction, device, and storage medium
CN114331728A (zh) 一种证券分析管理系统
Park et al. A new forecasting system using the latent dirichlet allocation (LDA) topic modeling technique
CN117786121B (zh) 一种基于人工智能的档案鉴定方法以及系统
CN117556118B (zh) 基于科研大数据预测的可视化推荐系统及方法
CN117829986A (zh) 一种金融贷款数字化信息管理系统架构
US20210042867A1 (en) System, Method and computer program product for evaluating admissions related data and compute comprehensible scores for further determination, evaluation, and conclusion of applicant-fit in a college or university system.

Legal Events

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

Ref document number: 22824229

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2022824229

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2022824229

Country of ref document: EP

Effective date: 20240118