CN114254201A - Recommendation method for science and technology project review experts - Google Patents
Recommendation method for science and technology project review experts Download PDFInfo
- Publication number
- CN114254201A CN114254201A CN202111587108.1A CN202111587108A CN114254201A CN 114254201 A CN114254201 A CN 114254201A CN 202111587108 A CN202111587108 A CN 202111587108A CN 114254201 A CN114254201 A CN 114254201A
- Authority
- CN
- China
- Prior art keywords
- expert
- recommendation
- model
- project
- candidate
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000005516 engineering process Methods 0.000 title claims abstract description 28
- 238000012552 review Methods 0.000 title claims abstract description 26
- 238000013178 mathematical model Methods 0.000 claims abstract description 23
- 238000004364 calculation method Methods 0.000 claims abstract description 18
- 238000010276 construction Methods 0.000 claims abstract description 14
- 238000011156 evaluation Methods 0.000 claims description 15
- 238000012216 screening Methods 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 7
- 238000011160 research Methods 0.000 claims description 7
- 238000005065 mining Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000000611 regression analysis Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 10
- 238000001914 filtration Methods 0.000 description 10
- 230000011218 segmentation Effects 0.000 description 8
- 238000000605 extraction Methods 0.000 description 7
- 230000008901 benefit Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000012827 research and development Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000009833 condensation Methods 0.000 description 1
- 230000005494 condensation Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000010076 replication Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
-
- 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
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- 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
- G06F16/958—Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/237—Lexical tools
- G06F40/247—Thesauruses; Synonyms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a recommendation method of science and technology project review experts, which comprises the following steps: reading an application form of a project to be evaluated and examined, and establishing a vector-based knowledge representation model of a project group to be evaluated and examined; reading the data of candidate experts in the basic library, and establishing a knowledge representation model of the project group to be evaluated and examined based on the object elements according to the characteristics of the expert information and the construction method of the expert object element knowledge representation model; calculating similarity values of the project groups and the candidate experts by adopting a similarity calculation method based on a knowledge representation model, and taking the similarity values as first recommended values of the candidate experts; respectively calculating the scores of the candidate experts on a preset index, and calculating a second recommendation value of the candidate experts according to a preset expert score mathematical model; and calculating the recommendation index of the candidate expert according to the first recommendation value and the second recommendation value to obtain a recommendation list of the recommendation order of the candidate expert according to the recommendation index. The method and the system can effectively improve the matching degree of the recommendation experts and the item content.
Description
Technical Field
The invention belongs to the technical field of power systems, and particularly relates to a recommendation method for science and technology project review experts.
Background
At present, the support policy of the science and technology projects in China guides the science and technology plans, special projects and the like, and meanwhile, different fund plans are respectively established by governments of various regions to support the development of the science and technology projects. The strong support of the national and local governments on scientific and technological activities directly leads to the increase of the number of the declaration and establishment of scientific and technological projects.
The prior art has the following defects and shortcomings: due to the fact that the accuracy and the scientificity of the mode determined by the scientific and technical project review expert are not enough, the phenomenon that the expert is not matched with the content of the evaluated project is common.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a recommendation method for a science and technology project review expert, so as to effectively improve the matching degree of the review expert and the science and technology project content.
In order to solve the technical problem, the invention provides a recommendation method for science and technology project review experts, which comprises the following steps:
step S1, reading the application books of the items to be evaluated, and establishing a vector-based knowledge representation model of the item group to be evaluated;
step S2, reading the data of the candidate experts in the basic library, and establishing a knowledge representation model of the project group to be evaluated and reviewed based on the object elements according to the characteristics of the expert information and the construction method of the expert object element knowledge representation model;
step S3, calculating the similarity value between the project group and the candidate expert by adopting a similarity calculation method based on a knowledge representation model, and taking the similarity value as a first recommendation value of the candidate expert;
step S4, respectively calculating the scores of the candidate experts on the preset indexes, and calculating second recommended values of the candidate experts according to a preset expert score mathematical model;
and step S5, calculating the recommendation index of the candidate expert according to the first recommendation value and the second recommendation value, and obtaining a recommendation list of the recommendation order of the candidate expert according to the recommendation index.
Further, the step S1 specifically includes: reading an application form of an item to be evaluated, firstly, adopting a Chinese keyword algorithm based on a word semantic network to calculate the key degree of the keyword and screening the keyword according to the key degree; then constructing a vector model according to a vector space model construction method and the mapping relation between the criticality and the keywords; and finally, according to the characteristic that the scientific and technological project is recommended by taking the group as a unit, adopting a merging strategy for the project model, and establishing a vector-based knowledge representation model of the project group to be evaluated.
Further, the step S2 specifically includes: reading expert data in a basic library, firstly adopting a Chinese keyword algorithm based on a word semantic network to calculate the key degree of the keywords and screening the keywords according to the key degree; then constructing a vector model according to a vector space model construction method and the mapping relation between the criticality and the keywords; and finally, establishing a knowledge representation model of the project group to be evaluated based on the object elements according to the characteristics of the expert information and the construction method of the expert object element knowledge representation model.
Further, the step S2 further includes: and establishing an expert mathematical model base according to the knowledge representation model of each expert, and establishing the knowledge representation model of the expert by reading database data.
Further, the step S3 specifically includes: firstly, respectively constructing conceptual hierarchical models of project groups and experts by adopting a hierarchical clustering algorithm, then calculating the similarity values of the science and technology project groups and the experts by adopting a node maximum depth improved cosine similarity calculation method, obtaining the first N candidate experts with the similarity values larger than a threshold value, and forming a preliminary recommendation list of the candidate experts; and taking the obtained similarity value as a first recommendation value of the candidate expert.
Further, the step S4 specifically includes: constructing an expert scoring mathematical model according to an expert evaluation analysis method and an expert evaluation principle, and establishing an expert evaluation system by adopting an analytic hierarchy process aiming at preset indexes of scientific research subjects, writings, titles and awards; respectively calculating scores of all preset indexes based on a regression analysis method; finally, constructing an expert scoring mathematical model based on an expert evaluation system and index scoring; and calculating a second recommended value of the candidate expert according to a preset expert scoring mathematical model.
Further, the step S4 further includes: and establishing an expert scoring model mathematical library according to the expert scoring model, and establishing the expert scoring mathematical model by reading the database data.
Further, the step S5 specifically includes: calculating the recommendation index of the candidate expert by the following formula:
S＝S1×M1+S2×M2
wherein S represents the final recommendation index, S1Represents a first recommended value, S2Represents a second recommended value; m1Weight representing the first recommended value, M2A weight representing the second recommendation value.
Further, the vector space model is established in the following way:
wherein, ti(i ═ 1, 2.. times, n) is a keyword entry, wi(d) Is tiWeight in d, key (W)i) Is WiA criticality value of;
and respectively processing the vector V according to different characteristics of the project and the expert information to form a knowledge representation model of the scientific project and a knowledge representation model of the expert.
Further, a knowledge representation model is constructed for the scientific and technical project and the expert information by adopting a text mining method.
The implementation of the invention has the following beneficial effects: by setting a keyword extraction method, knowledge model representation of scientific and technical projects and experts and a similarity method based on the knowledge model, potential and important relevance of the projects and the experts is analyzed by a text mining related method, so that the matching degree of the contents of the evaluation experts and the project to be evaluated is effectively improved;
by setting a recommendation algorithm based on content, a collaborative filtering recommendation algorithm, a knowledge-based recommendation algorithm, a recommendation algorithm based on association rules and a combined recommendation algorithm, the advantages, the disadvantages and the applicable occasions of each algorithm are analyzed by comparison, and a sufficient theoretical basis is provided for selecting which recommendation algorithm and how to realize in the realization of a scientific and technological project review expert recommendation system;
by setting a scientific and technological project review expert recommendation model for research, the matching degree of the scientific and technological projects and the expert contents is improved through the model according to a similarity calculation method of a scientific and technological project group and the experts; and then, the expert recommendation index is adjusted according to the weighting of the expert scoring mathematical model, so that the recommendation result is more scientific and effective.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart illustrating a recommendation method for a scientific and technological project review expert according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating the main steps of collaborative filtering recommendation according to an embodiment of the present invention.
FIG. 3 is a comparison diagram of a recommendation algorithm in an embodiment of the present invention.
Fig. 4 is a diagram of a model structure of a recommendation system commonly used in the embodiment of the present invention.
Fig. 5 is a diagram of a recommendation model of a science and technology project review expert in an embodiment of the invention.
FIG. 6 is a flowchart illustrating the design of a scientific project review expert recommendation system according to an embodiment of the present invention.
Fig. 7 is a structural diagram of a Lucene system in the embodiment of the present invention.
FIG. 8 is a diagram illustrating the establishment of a stop word lexicon in an embodiment of the present invention.
FIG. 9 is a diagram illustrating keyword extraction according to an embodiment of the present invention.
FIG. 10 is a conceptual model building and similarity calculation diagram according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments refers to the accompanying drawings, which are included to illustrate specific embodiments in which the invention may be practiced.
Referring to fig. 1, an embodiment of the present invention provides a method for recommending a scientific and technological project review expert, including:
step S1, reading the application books of the items to be evaluated, and establishing a vector-based knowledge representation model of the item group to be evaluated;
step S2, reading the data of the candidate experts in the basic library, and establishing a knowledge representation model of the project group to be evaluated and reviewed based on the object elements according to the characteristics of the expert information and the construction method of the expert object element knowledge representation model;
step S3, calculating the similarity value between the project group and the candidate expert by adopting a similarity calculation method based on a knowledge representation model, and taking the similarity value as a first recommendation value of the candidate expert;
step S4, respectively calculating the scores of the candidate experts on the preset indexes, and calculating second recommended values of the candidate experts according to a preset expert score mathematical model;
and step S5, calculating the recommendation index of the candidate expert according to the first recommendation value and the second recommendation value, and obtaining a recommendation list of the recommendation order of the candidate expert according to the recommendation index.
Specifically, referring to fig. 2-10, several recommendation algorithms are introduced, including a collaborative filtering recommendation algorithm, an association rule-based recommendation algorithm, a content-based recommendation algorithm, a knowledge-based recommendation algorithm, and a combined recommendation algorithm.
Recommendation algorithm for collaborative filtering
The collaborative filtering recommendation does not need to consider the characteristics and attributes of commodities but is recommended from the perspective of users, a system acquires recommendation information by learning implicit data such as records of commodities purchased by customers, browsing commodity records or grading commodities, the algorithm has the greatest advantage that no special requirement is required for a recommendation object, so that unstructured complex objects such as movies, music and other common commodities can be processed, the collaborative filtering recommendation avoids incomplete and inaccurate analysis based on content recommendation by sharing the experience of other people, and because the collaborative filtering recommendation does not depend on the characteristics and attributes of commodities, commodities which are different in surface characteristics but have great correlation in fact can be mined as recommendations, so that the users are helped to recommend new interesting commodities, potential but undiscovered interests of the users can be mined, and more importantly, the collaborative filtering recommendation can be carried out according to the records of commodities purchased by customers, And the knowledge system of the user is updated and increased by continuously accumulating implicit data such as browsing commodity records or grading commodities, and more information is provided for making more accurate recommendations later.
(II) recommendation algorithm based on association rule
The recommendation algorithm based on the association rule can be regarded as an inference technology to some extent, the recommendation algorithm is not established on the basis of user needs and preferences to generate recommendations, but utilizes specific rules formulated for specific fields to carry out inference based on the association rule and an example, the recommendation of the method is based on the association rule, purchased commodities are used as rule heads, recommendation objects are used as rule bodies, and the implementation steps for generating the recommendations are as follows:
the method comprises the following steps: finding out all association rules meeting the minimum support degree and the minimum confidence degree by using a recommendation algorithm of the association rules, and storing the association rules into a rule base R;
step two: setting a candidate recommendation set P for each current client C, and initializing to be empty;
step three: searching the rule base R to find out all association rule sets R supported by the customer C, namely all commodities on the left part of the association rules appear in the historical purchasing behavior record of the customer C;
step four: adding the commodity appearing at the right part of any rule in the set R into the candidate recommendation set P;
step five: deleting the commodities purchased by the user from the candidate recommendation set P;
step six: sorting all candidate items of the candidate recommendation set P from large to small according to the confidence degree of the association rule set R, and if one commodity appears in a plurality of rules, selecting the rule with the highest confidence degree as the most sorted standard;
step seven: selecting the top N items with the highest confidence coefficient from the candidate recommendation set P as recommendation results and returning the recommendation results to the client C;
the recommendation based on the association rule can find the mutual relevance of different commodities in the sales process, wherein the recommendation is successfully applied in the retail industry, online bookstores and electronic commerce, the management rule of the recommendation is that in a transaction database, the proportion of transactions purchasing a commodity set X is counted, and a commodity set Y is purchased at the same time, and the intuitive meaning is that the user tends to purchase other related commodities when purchasing some commodities.
(III) recommendation algorithm based on content
The content-based recommendation algorithm is to recommend other objects with similar attributes as recommendations according to object information selected by a user, is a continuation and development of an information filtering technology, is to build recommendations made on content text information of an item, does not need to build a vector space model according to evaluation opinions of the user on the item, and obtains information of interest materials of the user from text information related to feature description of the content by segmenting the user text information and building the vector space model, and the basic idea realized based on the content recommendation algorithm is as follows:
firstly, analyzing text information representing users by a characteristic extraction method for each user to obtain a data structure of an interest model capable of describing user interest material information; then, analyzing the text information representing the items by a characteristic extraction method for each item to obtain a data structure capable of describing the characteristics of the items; finally, when one user needs to be recommended, only the data structure of the user interest model representing the user needs to be compared with the feature vector matrixes of all items to obtain the similarity between the user and the items, and the system recommends the items according to the similarity value obtained through calculation.
The comparison of a collaborative filtering recommendation algorithm, a recommendation algorithm based on association rules, a recommendation algorithm based on content, a recommendation algorithm based on knowledge and a combined recommendation algorithm:
in the scientific and technological project review management, a large number of historical review records with high confidence degrees do not exist, project information and an expert information database do not have corresponding association rules, relatively speaking, an algorithm based on content recommendation is more suitable, and the recommendation method of the scientific and technological project review expert is provided in the embodiment aiming at the defect that potential interests cannot be found based on the content recommendation algorithm and the characteristics of scientific and technological projects. And (3) finishing the recommendation of the science and technology project review expert mainly by constructing a science and technology project and expert knowledge representation model, establishing an expert mathematical model base, an expert scoring mathematical model base, a reviewed project base and the like according to the information of the science and technology project review expert recommendation model structure diagram and displaying the structure in the science and technology project review expert recommendation model structure diagram, and then processing by using a recommendation algorithm according to the library information and related model data to generate a review expert recommendation list.
Referring to fig. 6, in step S1, reading an application form of a to-be-evaluated item, first, calculating the criticality of a keyword by using a chinese keyword algorithm based on a word semantic network, and screening the keyword accordingly; then constructing a vector model according to a vector space model construction method and the mapping relation between the criticality and the keywords; and finally, according to the characteristic that the scientific and technological projects are generally recommended in units of groups, adopting a merging strategy for the project models, and establishing a vector-based knowledge representation model of the project groups to be evaluated.
In step S2, reading the expert data in the basic library, firstly, adopting a Chinese keyword algorithm based on a word semantic network to calculate the key degree of the keywords and screening the keywords according to the key degree; then constructing a vector model according to a vector space model construction method and the mapping relation between the criticality and the keywords; and finally, establishing a knowledge representation model of the project group to be evaluated based on the object elements according to the characteristics of the expert information and the construction method of the expert object element knowledge representation model. In order to improve the operation efficiency of the recommendation system, an expert mathematical model base is established according to the knowledge representation model of each expert, and in the operation process of the recommendation algorithm, the knowledge representation model of the expert is established by reading database data.
Step S3 is to adopt an improved similarity calculation method based on the knowledge representation model to calculate the similarity between the project group and the expert after constructing the project group expert knowledge representation model through step S1 and step S2: the method comprises the steps of firstly, respectively constructing conceptual hierarchical models of a project group and experts by adopting a hierarchical clustering algorithm, then, calculating similarity values of the science and technology project group and the experts by adopting a node maximum depth improved cosine similarity calculation method, obtaining the first N candidate experts with the similarity values larger than a threshold value, and forming a preliminary recommendation list of the candidate experts. It is understood that the obtained similarity value is used as the first recommendation value of the candidate expert.
In the step S4, an expert evaluation mathematical model is constructed according to an expert evaluation analysis method and an expert evaluation principle, and an expert evaluation system is established by adopting an analytic hierarchy process aiming at preset indexes such as scientific research topics, works, titles, awards and the like; respectively calculating scores of all preset indexes based on a regression analysis method; and finally, providing an expert scoring mathematical model based on an expert evaluation system and index scoring. In order to improve the performance of the recommendation system, the embodiment of the invention establishes the expert rating model mathematical library according to the expert rating model, and in the operation process of the recommendation algorithm, the expert rating mathematical model is established by reading the database data. It is to be understood that the second recommendation value for the candidate expert is calculated according to an expert scoring mathematical model. As an example, the preset index scores may be summed to obtain the second recommendation value.
Step S5 compares the first recommended value and the second recommended value obtained in step S3 and step S4Processing recommended values, wherein the calculation formula is as follows: s ═ S1×M1+S2×M2Wherein S represents the final recommendation index, S1Represents a first recommended value, S2Represents a second recommended value; m1Weight representing the first recommended value, M2A weight representing the second recommendation value. And arranging the recommendation orders of the candidate experts according to the final recommendation index S and the size sequence to obtain a recommendation list of the candidate experts.
The experimental result shows that the recommendation result generated by applying the recommendation method has higher accuracy, and a system realized by applying the recommendation method has better feasibility.
In this embodiment, the similarity calculation method between the scientific and technical project and the expert is described as follows:
the main information sources of the scientific and technological project and the expert are database fields such as application books and expert histories, the fields are stored in a database in a semi-structured mode, and in order to better analyze the potential and important relevance of the project and the expert, the embodiment of the invention adopts a word segmentation technology, keyword extraction, knowledge representation and other text mining methods to construct a knowledge representation model for the scientific and technological project and the expert information;
word segmentation technology: the special situation of Chinese word segmentation, the Chinese words are different from English words, no obvious separation symbols exist between Chinese words, the Chinese words have various forming modes, the number of words forming a single word is different, furthermore, many characters in a sentence can be connected to describe a meaning, English is in units of words, the words are obviously separated and distinguished by spaces, the system only needs to divide according to the spaces, therefore, the word segmentation processing for Chinese character strings is more complex and difficult than English processing, and can be effectively searched by some special Chinese word segmentation processing methods, in the current common practical development and application, some improved Chinese word segmentation tools are added while the Lucene toolkit is used, practical application shows that the Lucene integrating the tools really achieves good effect;
it can be known from the structural system of Lucene that, no matter building indexes or participles, parsed texts need to pass through a parser, and then the texts are parsed into Token streams to be input to syntax parsing logic or index building logic of query statements, wherein the participle function is the most important indispensable part in the parser, and the Lucene system architecture mainly comprises the following three parts: the device comprises a basic packaging structure, an index core and an external interface, wherein the index core is the most important component;
extracting keywords: the keyword extraction method based on the word semantic network comprises the following steps: preprocessing a text based on a Chinese word segmentation method; mapping the scientific and technical project and the text of the expert into a word network based on the word semantic similarity; calculating the degree of intermediacy of the word network according to the concept of the social network; calculating the criticality according to the word medians and the statistical characteristics, screening key words according to the criticality to form a set, wherein the algorithm mainly comprises the following modules: text preprocessing, word semantic network construction, intervalidness calculation, criticality calculation and keyword screening;
the text mining method comprises the following steps: according to the requirement of relevant files determined by experts in scientific and technical projects, the following information is a main component for constructing a knowledge representation model:
item information:
(1) the project name and the title are a condensation point of the project information;
(2) the key technology and the public customs direction can indicate the specific research direction of the declaration project;
(3) the main research and development contents of the project are detailed descriptions of the specific mode and content of the research of the declaration project and the expected results which can be achieved;
(4) the project main technical indexes and economic indexes reflect the project plan target and the actual situation;
(5) the feasibility report is used for reporting various aspects of environment, policy, law and the like from economy, technology, research and development, operation to society of a unit to which the project belongs, researching, analyzing and discussing, forecasting various interest factors and feasibility of the project, and estimating indexes such as project risk, economic contribution, social benefit and the like;
expert information:
(1) familiarity with the specialty, the research specialty that the expert is engaged in;
(2) the direction of study, the specific direction studied by the expert;
(3) resume of the expert, personal image of the expert, including written representation of seniority and competency;
(4) various awards obtained;
(5) journal publishing conditions;
(6) the task is to complete the situation.
The method comprises the following steps of representing the scientific and technical project and knowledge of experts:
VSM has strong expression and expansion capability, and is used as the simplest and most effective knowledge representation model, the model is widely applied in various fields, including the concept and method of an object model proposed later, which are also based on the result of expansion of a vector space model, and at present, many related fields including text replication detection, full text retrieval and the like are applied to the technologies of feature item selection, feature weighting strategy and the like in the vector space model;
the modeling of the scientific and technological project to be reviewed and the review expert information adopts a knowledge representation method, wherein the basic idea of establishing a vector space model in the system is to perform key words and key value after the text information of an input object is processed by algorithms such as word segmentation, key word extraction and the like, and after the key value is normalized, a vector space model can be established through the following steps:
V(d)＝{＜t1，w1(d)＞，＜t2，w2(d)＞，...，＜tn，wn(d)＞}，
wherein t isi(i ═ 1, 2.. times, n) is a keyword entry, wi(d) Is tiWeight in d, key (W)i) Is WiA criticality value of;
and respectively processing the vector V according to different characteristics of the project and the expert information to form a knowledge representation model V of the scientific project and a knowledge representation model V of the expert.
The similarity measurement method based on the knowledge representation model is explained as follows:
for the synonymous relation, the concept context relation and the like between the keywords in the scientific and technological project group and the expert knowledge representation and other keywords, the similarity detection can not be effectively carried out on the documents by adopting a mode of directly inquiring and matching the knowledge model, and in order to improve the accuracy of the recommendation algorithm, the similarity calculation is carried out after hierarchical clustering is carried out on the keywords of the knowledge model;
the recommendation model mainly consists of three steps: clustering knowledge representations based on semantic similarity and constructing a concept tree model of each project group and an expert concept tree model set; calculating the similarity value between the project group concept tree and each expert concept tree by adopting a cosine similarity method improved by maximum depth matching of nodes; and giving the recommendation order of the project group review experts according to the similarity value, and outputting a recommendation expert list.
Similarity detection can not be effectively carried out on the documents by adopting a mode of directly inquiring and matching a knowledge model through synonymy relations, concept context relations and the like between the scientific and technological project group and other keywords possibly existing in the keywords in the expert knowledge representation, and in order to improve the accuracy of a recommendation algorithm, similarity calculation is carried out after hierarchical clustering is carried out on the keywords of the knowledge model; the recommendation model mainly consists of three steps: clustering knowledge representations based on semantic similarity and constructing a concept tree model of each project group and an expert concept tree model set; calculating the similarity value between the project group concept tree and each expert concept tree by adopting a cosine similarity method improved by maximum depth matching of nodes; and giving the recommendation order of the project group review experts according to the similarity value, and outputting a recommendation expert list.
As can be seen from the above description, the present invention provides the following advantageous effects: potential and important relevance of the project and the expert is analyzed through a relevant text mining method, so that the matching degree of the recommended expert and the project content is effectively improved; the advantages, the disadvantages and the applicable occasions of all algorithms are contrastively analyzed, so that a sufficient theoretical basis is provided for selecting which recommendation algorithm and how to realize in the realization of the scientific and technological project review expert recommendation system; according to the model, firstly, the matching degree of science and technology projects and expert contents is improved according to a similarity calculation method of a science and technology project group and experts; and then, the expert recommendation index is adjusted according to the weighting of the expert scoring mathematical model, so that the recommendation result is more scientific and effective.
The above disclosure is only for the purpose of illustrating the preferred embodiments of the present invention, and it is therefore to be understood that the invention is not limited by the scope of the appended claims.
Claims (10)
1. A recommendation method for science and technology project review experts is characterized by comprising the following steps:
step S1, reading the application books of the items to be evaluated, and establishing a vector-based knowledge representation model of the item group to be evaluated;
step S2, reading the data of the candidate experts in the basic library, and establishing a knowledge representation model of the project group to be evaluated and reviewed based on the object elements according to the characteristics of the expert information and the construction method of the expert object element knowledge representation model;
step S3, calculating the similarity value between the project group and the candidate expert by adopting a similarity calculation method based on a knowledge representation model, and taking the similarity value as a first recommendation value of the candidate expert;
step S4, respectively calculating the scores of the candidate experts on the preset indexes, and calculating second recommended values of the candidate experts according to a preset expert score mathematical model;
and step S5, calculating the recommendation index of the candidate expert according to the first recommendation value and the second recommendation value, and obtaining a recommendation list of the recommendation order of the candidate expert according to the recommendation index.
2. The recommendation method according to claim 1, wherein the step S1 specifically includes: reading an application form of an item to be evaluated, firstly, adopting a Chinese keyword algorithm based on a word semantic network to calculate the key degree of the keyword and screening the keyword according to the key degree; then constructing a vector model according to a vector space model construction method and the mapping relation between the criticality and the keywords; and finally, according to the characteristic that the scientific and technological project is recommended by taking the group as a unit, adopting a merging strategy for the project model, and establishing a vector-based knowledge representation model of the project group to be evaluated.
3. The recommendation method according to claim 2, wherein the step S2 specifically includes: reading expert data in a basic library, firstly adopting a Chinese keyword algorithm based on a word semantic network to calculate the key degree of the keywords and screening the keywords according to the key degree; then constructing a vector model according to a vector space model construction method and the mapping relation between the criticality and the keywords; and finally, establishing a knowledge representation model of the project group to be evaluated based on the object elements according to the characteristics of the expert information and the construction method of the expert object element knowledge representation model.
4. The recommendation method according to claim 3, wherein the step S2 further comprises: and establishing an expert mathematical model base according to the knowledge representation model of each expert, and establishing the knowledge representation model of the expert by reading database data.
5. The recommendation method according to claim 4, wherein the step S3 specifically comprises: firstly, respectively constructing conceptual hierarchical models of project groups and experts by adopting a hierarchical clustering algorithm, then calculating the similarity values of the science and technology project groups and the experts by adopting a node maximum depth improved cosine similarity calculation method, obtaining the first N candidate experts with the similarity values larger than a threshold value, and forming a preliminary recommendation list of the candidate experts; and taking the obtained similarity value as a first recommendation value of the candidate expert.
6. The recommendation method according to claim 5, wherein the step S4 specifically comprises: constructing an expert scoring mathematical model according to an expert evaluation analysis method and an expert evaluation principle, and establishing an expert evaluation system by adopting an analytic hierarchy process aiming at preset indexes of scientific research subjects, writings, titles and awards; respectively calculating scores of all preset indexes based on a regression analysis method; finally, constructing an expert scoring mathematical model based on an expert evaluation system and index scoring; and calculating a second recommended value of the candidate expert according to a preset expert scoring mathematical model.
7. The recommendation method according to claim 6, wherein the step S4 further comprises: and establishing an expert scoring model mathematical library according to the expert scoring model, and establishing the expert scoring mathematical model by reading the database data.
8. The recommendation method according to claim 7, wherein the step S5 specifically includes: calculating the recommendation index of the candidate expert by the following formula:
S＝S1×M1+S2×M2
wherein S represents the final recommendation index, S1Represents a first recommended value, S2Represents a second recommended value; m1Weight representing the first recommended value, M2A weight representing the second recommendation value.
9. The recommendation method of claim 1, wherein the vector space model is established by:
V(d)＝{<t1，w1(d)>，<t2，w2(d)＞，…，＜tn，wn(d)>}，
wherein, ti(i ═ 1, 2.. times, n) is a keyword entry, wi(d) Is tiWeight in d, key (W)i) Is WiA criticality value of;
and respectively processing the vector V according to different characteristics of the project and the expert information to form a knowledge representation model of the scientific project and a knowledge representation model of the expert.
10. The recommendation method according to claim 1, wherein the knowledge representation model is constructed by text mining for the scientific and technical items and the expert information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111587108.1A CN114254201A (en) | 2021-12-23 | 2021-12-23 | Recommendation method for science and technology project review experts |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111587108.1A CN114254201A (en) | 2021-12-23 | 2021-12-23 | Recommendation method for science and technology project review experts |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114254201A true CN114254201A (en) | 2022-03-29 |
Family
ID=80796936
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111587108.1A Pending CN114254201A (en) | 2021-12-23 | 2021-12-23 | Recommendation method for science and technology project review experts |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114254201A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114282976A (en) * | 2021-12-27 | 2022-04-05 | 赛尔网络有限公司 | Supplier recommendation method and device, electronic equipment and medium |
| CN115879829A (en) * | 2023-02-21 | 2023-03-31 | 广东省科技基础条件平台中心 | Evaluation expert screening method applied to platform innovation capability examination and verification |
| CN116129356A (en) * | 2023-02-02 | 2023-05-16 | 南通市亿控自动化系统有限公司 | Monitoring data analysis method and system |
| CN117093670A (en) * | 2023-07-18 | 2023-11-21 | 北京智信佳科技有限公司 | Method for realizing intelligent recommending expert in paper |
| CN117131279A (en) * | 2023-09-13 | 2023-11-28 | 合肥工业大学 | Data processing method and device for expert recommendation |
| CN117235373A (en) * | 2023-11-14 | 2023-12-15 | 四川省计算机研究院 | Scientific research hot spot recommendation method based on information entropy |
| WO2024164697A1 (en) * | 2023-02-07 | 2024-08-15 | 中国计量科学研究院 | Method and apparatus for recommending test organization in scientific and technical instrument/device testing |
-
2021
- 2021-12-23 CN CN202111587108.1A patent/CN114254201A/en active Pending
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114282976A (en) * | 2021-12-27 | 2022-04-05 | 赛尔网络有限公司 | Supplier recommendation method and device, electronic equipment and medium |
| CN116129356A (en) * | 2023-02-02 | 2023-05-16 | 南通市亿控自动化系统有限公司 | Monitoring data analysis method and system |
| CN116129356B (en) * | 2023-02-02 | 2023-10-24 | 南通市亿控自动化系统有限公司 | Monitoring data analysis method and system |
| WO2024164697A1 (en) * | 2023-02-07 | 2024-08-15 | 中国计量科学研究院 | Method and apparatus for recommending test organization in scientific and technical instrument/device testing |
| CN115879829A (en) * | 2023-02-21 | 2023-03-31 | 广东省科技基础条件平台中心 | Evaluation expert screening method applied to platform innovation capability examination and verification |
| CN117093670A (en) * | 2023-07-18 | 2023-11-21 | 北京智信佳科技有限公司 | Method for realizing intelligent recommending expert in paper |
| CN117131279A (en) * | 2023-09-13 | 2023-11-28 | 合肥工业大学 | Data processing method and device for expert recommendation |
| CN117235373A (en) * | 2023-11-14 | 2023-12-15 | 四川省计算机研究院 | Scientific research hot spot recommendation method based on information entropy |
| CN117235373B (en) * | 2023-11-14 | 2024-03-15 | 四川省计算机研究院 | Scientific research hot spot recommendation method based on information entropy |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | A content-based recommender system for computer science publications | |
| CN108491377B (en) | E-commerce product comprehensive scoring method based on multi-dimensional information fusion | |
| US11663254B2 (en) | System and engine for seeded clustering of news events | |
| CN114254201A (en) | Recommendation method for science and technology project review experts | |
| Avasthi et al. | Techniques, applications, and issues in mining large-scale text databases | |
| CN111061939B (en) | Scientific research academic news keyword matching recommendation method based on deep learning | |
| Li et al. | An intelligent approach to data extraction and task identification for process mining | |
| Baishya et al. | SAFER: sentiment analysis-based fake review detection in e-commerce using deep learning | |
| Jonathan et al. | Sentiment analysis of customer reviews in zomato bangalore restaurants using random forest classifier | |
| Yoon et al. | Doc2vec-based link prediction approach using SAO structures: Application to patent network | |
| Sheikhattar et al. | A thematic analysis–based model for identifying the impacts of natural crises on a supply chain for service integrity: A text analysis approach | |
| CN113538106A (en) | Commodity refinement recommendation method based on comment integration mining | |
| Zadgaonkar et al. | An Approach for analyzing unstructured text data using topic modeling techniques for efficient information extraction | |
| Das et al. | A review on text analytics process with a CV parser model | |
| CN116823321A (en) | Method and system for analyzing economic management data of electric business | |
| Maylawati et al. | Implicit aspect extraction in product reviews using FIN algorithm | |
| Hamad et al. | Sentiment analysis of restaurant reviews in social media using naïve bayes | |
| Zhang et al. | A text mining based method for policy recommendation | |
| Amira et al. | Opinion Analysis of Traveler Based on Tourism Site Review Using Sentiment Analysis | |
| Chen et al. | A template approach for summarizing restaurant reviews | |
| CN113869038A (en) | Attention point similarity analysis method for Baidu stick bar based on feature word analysis | |
| Chen et al. | Towards accurate search for e-commerce in steel industry: a knowledge-graph-based approach | |
| Pawade et al. | Survey on Resume and Job Profile Matching System | |
| Pandi et al. | Reputation based online product recommendations | |
| Urkude et al. | Comparative analysis on machine learning techniques: a case study on Amazon product |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |