CN111078873A - Domain expert selection method based on citation network and scientific research cooperation network - Google Patents
Domain expert selection method based on citation network and scientific research cooperation network Download PDFInfo
- Publication number
- CN111078873A CN111078873A CN201911154794.6A CN201911154794A CN111078873A CN 111078873 A CN111078873 A CN 111078873A CN 201911154794 A CN201911154794 A CN 201911154794A CN 111078873 A CN111078873 A CN 111078873A
- Authority
- CN
- China
- Prior art keywords
- network
- scholars
- student
- citation
- nodes
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Computational Linguistics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method for selecting domain experts based on a citation network and a scientific research cooperation network. The method comprises the following steps: firstly, constructing a student cooperative directed network model based on database metadata information to generate a student cooperative network; secondly, constructing a document reference model based on document reference information, deleting self-induced interference, and generating a learner reference network by linear mapping; secondly, fusing the student cooperation network and the student reference network to generate a student relation network; and finally, calculating important nodes in the scholars relationship network and dividing the community structure, wherein the result is used as a selection expert list. According to academic activities and academic achievements of the scholars, the two standards of academic ability evaluation and cooperative network quality evaluation are comprehensively considered, so that corresponding experts can be quickly and accurately recommended, the field where the experts are good can be identified, and the problem that professional matching of the experts is inaccurate in the conventional evaluation and selection method is solved; the importance of the senior scholars in scientific research work is visually embodied, and the selected experts are more reasonable.
Description
Technical Field
The invention belongs to the technical field of big data document retrieval, and particularly relates to a method for selecting domain experts based on a citation network and a scientific research cooperation network.
Background
With the popularization of computing technology and the rapid development of network informatization, the selection work of science and technology experts is developed into a network intelligent mode from a manual mode. The network intelligent expert selection mode fundamentally solves the problems of low efficiency, lack of scientificity, impartiality and the like of the traditional selection mode, breaks through the limitation of regions of experts in the traditional selection mode, expands the core of the expert team network intelligent expert selection mode into an expert selection algorithm, and is mainly classified into two types based on text analysis and network structure at present.
Based on the text analysis of research topics, contents, disciplines and the like, experts are screened by combining similarity calculation, mathematical model counting and the like. The Chinese patent with the application number of 201410092584.X describes a subject characteristic value algorithm and a project review expert recommendation algorithm based on the same. And calculating the text similarity of the research content of the project and the research direction of the evaluation expert according to the text information, calculating subject characteristic values of the project and the evaluation expert according to the national standard subject classification and code, and finishing recommendation work by combining the text similarity and the subject characteristic values of the project and the evaluation expert. On the basis of calculating the similarity of the text feature vectors, scientific and reasonable subject classification standards are added, accurate calculation is performed in a subject level, the proportion of each level of subject can be fully considered, and the subdivision degree of the subject is emphasized. However, with the development of science and technology, the research of cross disciplines or interdisciplinary research is increased, and the experts which continue to use the existing disciplinary classification calculation selection inevitably bring the deviation of the result.
An expert selection method based on a literature and student network structure mainly judges the academic influence of an expert by utilizing a link structure of a constructed network. Chinese patent application No. 201811228086.8 discloses a collaborative recommendation method based on expert domain similarity and association relationship. Taking batch thesis data as a training set, constructing a cooperative relationship network, calculating the shortest path between scholars by using a Dijkstra algorithm to serve as an expert correlation COR, constructing an expert word vector model by using a word2vec algorithm to calculate the cosine similarity between a correlation expert word vector and a field word vector to serve as the expert field similarity, and screening the expert with the expert field similarity SIM and the expert correlation COR meeting a threshold value to obtain a recommended expert. The expert association degree provided by the method is calculated according to the cooperative relationship between the experts, and the recommended experts are closely associated with the given experts according to the cooperative relationship. But only considering that the cooperative relationship is easily influenced by subjective factors, the objective limit standard of the research field is lacked; moreover, the cooperation between scholars cannot reflect the implicit relevance between the inheritance of knowledge and research topics.
The present invention has been made in view of this situation.
Disclosure of Invention
The technical problem to be solved by the invention is to overcome the defects of the prior art, and provide a method for selecting domain experts based on a citation network and a scientific research cooperation network, wherein two standards of academic ability evaluation and cooperation network quality evaluation are comprehensively considered, and the domain experts are quickly and accurately selected.
In order to solve the technical problems, the invention adopts the technical scheme that: a method for realizing field expert selection based on a citation network and a scientific research cooperation network by fusing and generating a scholars relationship network and borrowing important nodes and community division algorithms in a complex network comprises the following steps:
firstly, constructing a student cooperative directed network model based on database metadata information to generate a student cooperative network;
secondly, constructing a document reference model based on document reference information, deleting self-induced interference, and generating a learner reference network by linear mapping;
then, fusing the student cooperation network and the student reference network to generate a student relation network;
finally, calculating important nodes in the scholars relationship network and clustering and grouping, wherein the result is a selected expert list;
wherein the student cooperative network is a directed network.
In the scheme, the directed network has the characteristic of small clustering coefficient, so that the relation among scholars is clearer. The characteristic paths among all nodes of the directed network are short, namely learners can mutually know through short intermediate learners, and compared with the undirected network, the directed network is more accurate and efficient in discovering partners in the scientific research field. In addition, on the basis of the directed network, social relationships such as teachers and students and friends can be further integrated into the student cooperative directed network as node attributes, and clearer social relationship display among students can be obtained so as to better evaluate scientific research teams.
The further scheme of the invention is as follows: the establishment of the student cooperative network comprises the following steps: gcollaorati on=(V,Ecollaborat ion),Gcollaorati on=(V,Ecollaborat ion) Is composed of N scholar nodes and Ecollaboration|=McollaborationA directed network of edges, wherein GcollborationRepresents a set of scholars and scholars' cooperative relations in a scholars cooperative network, V represents a scholars group GcollborationSet of trainees in (E)collaborationRepresentative group of scholars GcollaborationThe cooperative relationship among the Chinese scholars;representative of scholars viAs a collaboration center, scholars viAnd scholars vjThe cooperative relationship of (1); in-scholar partnership collection GcollborationMiddle school student viAs a collaboration center, scholars viThe total number of times of cooperation with other scholars isIf the scholar viAnd scholars vjThere is a common achievement of scientific research, and the scholars viAs the scholars of the research result cooperation center, scholars viAnd scholars vjThere is a cooperation between them, denoted 1,is a scholar viAnd scholars vjSum of number of edges, i.e. by scholar viAs a collaboration center, scholars viAnd scholars viSum of the number of collaborations of; otherwise, the learner viAnd scholars vjThere is no cooperative relationship between them, and it is marked as 0; wherein i is more than or equal to 1, j is more than or equal to 1,andthe same or different;
the cooperation center scholars can calculate distribution through a computing terminal and can also set manually;
preferably, a first person in charge of scientific and technological projects, a communicator of journal papers, a communicator of conference papers, a first contributor of newspaper reports, a first inventor of patents, a first main edition of writings, a first standard drafter and a first reporter of research reports are selected as cooperation center scholars; if the journal paper does not indicate a communication student, the first student of the journal paper is used as a cooperation center student; if the conference paper does not indicate a communication student, the first student of the periodical paper is used as a cooperation center student;
more preferably, the first person in charge of the scientific research is taken as a scholarer of the cooperation center.
In the scheme, each student in the scientific research results is a node in the student cooperation network, the same student in the scientific research results is only used as a node, and multiple students in the same scientific research results have a cooperation relationship with each other; the scientific achievements include scientific projects, journal papers, meeting papers, newspaper reports, patents, writings, standards and research reports.
The further scheme of the invention is as follows: the learner reference network is constructed as follows: gcitation=(V,Ecitation),Gcitation=(V,Ecitation) Is composed of N scholar nodes and Ecitation|=McitationA directed network formed by edges; wherein G iscitationRepresentative scholars quote scholars in network and scholars quote customsSet of lines, V represents the group of scholars GcitationSet of trainees in (E)citationRepresentative group of scholars GcitationReference relations between the Chinese scholars; ,representative scholars quote scholars v in networkiAnd scholars vjIn the student, the reference relationship set GcitationChinese scholar viBy other scholars vjThe number of references totals up toIf the scholar viV. a certain scientific research result in scholarsjIf the scientific research result is quoted, the number is marked as 1,is a scholar viAnd scholars vjSum of directed edges, i.e. scholar viIs learnt of scientific achievements vjThe sum of the number of references of (a); if the scholar viHas no learners vjIf the reference is made, marking as 0; wherein i is more than or equal to 1, j is more than or equal to 1,andthe same or different.
In the scheme, the student node distribution of the student reference network and the student cooperative network is the same, and the difference is that the edges representing the node relationship in the student reference network and the student cooperative network are different.
The further scheme of the invention is as follows: the student relationship network is formed by fusing a student cooperation network and a student reference network:
G=α·Gcollaboration+β·Gcitation;
α and β are assigned values for weights of the student cooperative network and the student reference network, the α + β is 1, and the assignments of α and β can be calculated and assigned through a computing terminal or can be set manually;
the scholars relationship model G ═ (V, E)
the scholars relationship network is a directed network consisting of | V | ═ N scholars nodes and | E | ═ M edges; wherein G represents the set of scholars and relationships among scholars in the scholars group, V represents the set of scholars in the scholars group G, E represents the relationships among scholars in the scholars group G, EijStudent v in representative student relationship networkiAnd scholars vjIn the context of (a) or (b),is a node viThe strength of the penetration of (A) into (B),is a node viThe strength of (2).
The further scheme of the invention is as follows: the method for calculating the important nodes in the student relationship network comprises the steps of sequencing the importance of students in the student cooperation network according to an importance sequencing method of the nodes in the complex network;
preferably, the method for ranking according to the importance of the nodes in the complex network includes: the scholars are sorted by a method of recursively peeling off the network based on node strength, which specifically comprises the following steps:
s11, setting k as an integer, and taking k as 0;
s12, removing all student nodes with the strength not greater than k in the student relationship network, and simultaneously deleting edges connected with the student nodes with the strength not greater than k;
s13, checking the output intensity of the rest student nodes in the network at the moment, judging whether the student nodes with the intensity not greater than k still exist in the rest student nodes, if so, executing a step S12, otherwise, executing a step S14;
s14, forming a kth layer by the scholar nodes with the intensity not greater than k removed in the step S12, endowing the nodes of the kth layer with a node Ks value, namely equal to k, and then executing a step S15;
s15, judging whether the number of the nodes left in the network is 0, if so, executing a step S17, otherwise, executing a step S16
S16, setting k to k +1, and repeatedly executing step S12;
s17, sorting the scholars in a descending order according to the Ks values.
In the above scheme, in the nodes of each layer formed in step S14, the students of the same layer have the same Ks value, and the bit order of the students with the same Ks value in the permutation is the same.
The further scheme of the invention is as follows: the clustering grouping comprises community division of a scholar relationship network, the divided scholar communities are regarded as relatively independent research fields, and the method comprises the following steps:
s21, taking the scholar relationship network G as an initial network and setting the initial network as a current network;
s22, randomly dividing the nodes in the current network into two communities, and then executing a step S23;
s23, calculating the contribution degree of each node to the modularity degree, calculating the network modularity degree according to the contribution degree, and then executing the step S24;
s24, moving the nodes with lower contribution degree from one community to another community, and then executing the step S25;
s25, recalculating the contribution degree of each node to the modularity and the network modularity, and then executing the step S26;
s26, judging whether the network modularity is increased or not, simultaneously judging whether the network modularity reaches the maximum value or not, if the network modularity is increased, namely the maximum value is not reached, keeping the moving result of the node and returning to the step S24, and if the network modularity is not increased, withdrawing the mobile node, moving a new node with lower contribution degree different from the withdrawn mobile node from one community to another community, and returning to the step S25; if the modularity reaches the maximum value, executing step S27;
s27, recording and storing the network modularity and community structure of the initial network at the moment, and then executing a step S28;
and S28, continuously dividing each community divided in the step S27 as an individual network, and executing the steps S22-S28 for each individual network in a recursive mode until no larger modularity is generated in the initial network, so as to obtain a network community division result.
The further scheme of the invention is as follows: the contribution degree lambda of each node to modularity degreeiCalculated according to the following formula:
wherein, κr(i)Representing nodes v belonging to a community riThe sum of the edge values representing the relationship with other nodes in the community,is a node viThe strength of the penetration of (A) into (B),is a node viA strength ofr(i)Representing the ratio of the variable values of the nodes in the community r.
The further scheme of the invention is as follows: the modularity Q is calculated according to the following equation:
wherein m is the sum of the edge values representing the relationship in the scholars relationship network.
One embodiment of the selection method comprises the following steps: and carrying out community division on the student relationship network G obtained by fusion, wherein the division result is a plurality of student communities corresponding to each sub-technical field under the technical subject to which the student belongs, on the basis, node importance arrangement calculation is respectively carried out on each sub-field student relationship network obtained by division (namely a plurality of new student relationship networks divided by the student relationship network G), and an expert list is formed by the students with higher Ks values in each sub-field student relationship network.
In the above scheme, when the research field included in a specific technical subject is numerous and complicated, if a method of directly sorting the importance of nodes in the student relationship network G to obtain an expert list is adopted, the experts in the hot sub-technical field are ranked higher, and the experts in the cold sub-technical field are not selected into the list due to the ranking, which may result in the loss of some experts. Aiming at the problem, the invention preferentially adopts clustering grouping to divide the sub-fields of the student relationship network G, and then carries out importance sequencing, thereby effectively avoiding missing expert students in the cold field.
Another embodiment of the selection method comprises: and performing node importance arrangement calculation on the student relationship network G obtained by fusion, selecting student nodes with higher Ks values in the student relationship network G to form a new student relationship network, then performing community division on the newly formed student relationship network, wherein the division result is a plurality of student communities corresponding to each sub-technical field in the technical field to which the student belongs, and forming an expert list according to each sub-field.
In the above scheme, when the research field branches included in the specific technical subject are simple, if the learner relationship network G is clustered and grouped directly, a large number of expert learners may exist in each sub-field, and the lead learners in the research field cannot be obtained intuitively in the list. In addition, when identifying authoritative experts in a specific technical subject who are in a research hotspot, the calculation process of clustering first and then sorting learners is increased by meaningless calculation amount compared with the process of sorting first and then clustering. Aiming at the problems, the importance of the student nodes in the student relational network is preferentially sorted, the threshold value can be set to determine the student nodes with higher Ks so as to form a new student relational network, and then clustering grouping is carried out, so that a more visual expert list can be obtained, and more important expert and students in research hotspots in each sub-field are embodied.
The further scheme of the invention is as follows: the method for constructing the document reference model, deleting self-introduced interference and generating the student reference network by linear mapping based on the document reference information comprises the following steps of:
s31, constructing a document citation network model, counting the citation times and the other citation times of the document, and then executing the step S32;
and S32, generating a self-introduced document other-introduced network model according to the document other-introduced times mapping, and mapping to generate a student cited network.
Among the above schemes, the ith reference in S31 is in reference group GliteratureBy referencing variablesIs obtained by accumulation, and the calculation formula isThe ith literature in step S31 is in literature group GliteratureBy referencing a variableAnd coefficient of self-inductionProduct of (2)Is obtained by accumulation, and the calculation formula isIf the jth document cites the ith document, thenEqual to 1; if the jth document does not refer to the ith document, thenEqual to 0; if there is at least one scholar identity between the ith and jth documents, and the citation is self-citation, thenEqual to 0; if the ith article and the jth article do not have the same scholars, and the citation is other citations, thenEqual to 1; wherein i is more than or equal to 1, and j is more than or equal to 1; in the step S32, the number of times of the reference is determinedGenerating a self-introduced excluded document tare network model Gliterature-citedThen generates the learner reference network G as described above by linear mappingcitation。
The further scheme of the invention is as follows: the selection method comprises the steps of extracting student information in database metadata information and carrying out data cleaning, wherein in the data cleaning step, "student name + primary mechanism + secondary mechanism" is used as a unique identifier of each student and is endowed with an ID number.
After adopting the technical scheme, compared with the prior art, the invention has the following beneficial effects:
1. the mutual reference of the scholars reflects the reference influence of the scholars in the knowledge network, and the cooperative relationship of the scholars reflects the cooperative influence of the scholars in the social network. The method has the advantages that the relationship network of the scholars is constructed by the fusion of the citation network and the scientific research cooperation network, the potential association relationship existing between the scholars is disclosed by comprehensively considering the correlation of the research content and the influence of the experts, and the distribution condition of the field experts is reflected more comprehensively, objectively and accurately.
2. By constructing a student relationship network selection expert, the limitation of study subjects and study fields of students is broken through, and the effectiveness and reliability of clustering calculation of the students can be improved by fully utilizing the characteristic information of cooperation and reference relationship among the students; communities are divided based on the student relationship network, students in the same community often have the same or similar academic research fields, and the method is also suitable for experts recommending cross disciplines or emerging disciplines.
3. The cooperative contribution degree of the scholars is distinguished, the cooperative directed network of the scholars is constructed, the leading and cooperation effects among the cooperative scholars are disclosed, and the real core scholars are identified more conveniently. Based on the expert selection model algorithm which highlights the contribution degree of the core scholars, the recommended experts have more power.
The following describes embodiments of the present invention in further detail with reference to the accompanying drawings.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention, are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention without limiting the invention to the right. It is obvious that the drawings in the following description are only some embodiments, and that for a person skilled in the art, other drawings can be derived from them without inventive effort. In the drawings:
FIG. 1 is a flow chart of an expert selection method of the present invention;
FIG. 2 is a schematic diagram of the expert selection method of the present invention;
FIG. 3 is a schematic flow chart of the method for recursively peeling off the network based on node output strength to rank scholars according to the present invention;
FIG. 4 is a schematic diagram of the process of clustering and grouping scholars based on a scholars relationship network according to the present invention;
FIG. 5 is a schematic diagram of a model for constructing a student cooperation network and a student reference network in the invention;
FIG. 6 is a schematic diagram of a converged-generative student relationship network model in the present invention.
It should be noted that the drawings and the description are not intended to limit the scope of the inventive concept in any way, but to illustrate it by a person skilled in the art with reference to specific embodiments.
Detailed Description
As shown in fig. 1 to 6, the invention introduces a method for realizing domain expert selection by fusing and generating a scholars relationship network and borrowing important nodes and community division algorithms in a complex network based on a citation network and a scientific research cooperation network. Firstly, constructing a student cooperative directed network model based on database metadata information to generate a student cooperative network; secondly, constructing a document reference model based on document reference information, deleting self-induced interference, and generating a learner reference network by linear mapping; then, fusing the student cooperation network and the student reference network to generate a student relation network; and finally, calculating important nodes in the scholars relationship network and clustering and grouping, wherein the result is the selected expert list.
Examples
As shown in fig. 1, the present embodiment specifically includes the following steps:
A. constructing a student cooperative directed network model based on database metadata information, and generating a student cooperative network, wherein the student cooperative network is a directed network;
in this embodiment, the learner cooperative network in step a is: gcollaorati on=(V,Ecollaborat ion),Gcollaorati on=(V,Ecollaborat ion) Is composed of N scholar nodes and Ecollaboration|=McollaborationA directed network of edges, wherein GcollborationRepresents a set of scholars and scholars' cooperative relations in a scholars cooperative network, V represents a scholars group GcollborationSet of trainees in (E)collaborationRepresentative group of scholars GcollaborationThe cooperative relationship among the Chinese scholars;representative of scholars viAs a collaboration center, scholars viAnd scholars vjThe cooperative relationship of (1); in-scholar partnership collection GcollborationMiddle school student viAs a collaboration center, scholars viThe total number of times of cooperation with other scholars isIf the scholar viAnd scholars vjThere is a common achievement of scientific research, and the scholars viAs the scholars of the research result cooperation center, scholars viAnd scholars vjThere is a cooperation between them, denoted 1,is a scholar viAnd scholars vjThe number of edgesAnd, i.e. with the scholars viAs a collaboration center, scholars viAnd scholars viSum of the number of collaborations of; otherwise, the learner viAnd scholars vjThere is no cooperative relationship between them, and it is marked as 0; wherein i is more than or equal to 1, j is more than or equal to 1,andthe same or different.
In this embodiment, the cooperation center scholars may calculate and distribute through a computing terminal, or may set manually; preferably, a first person in charge of scientific and technological projects, a communicator of journal papers, a communicator of conference papers, a first contributor of newspaper reports, a first inventor of patents, a first main edition of writings, a first standard drafter and a first reporter of research reports are selected as cooperation center scholars; if the journal paper does not indicate a communication student, the first student of the journal paper is used as a cooperation center student; if the conference paper does not indicate a communication student, the first student of the periodical paper is used as a cooperation center student; more preferably, the first person in charge of the scientific research is taken as a scholarer of the cooperation center.
In this embodiment, each student in the scientific research result is a node in the student cooperation network, the same student in the scientific research results is only a node, and multiple students in the same scientific research result have a cooperative relationship with each other; the scientific achievements include scientific projects, journal papers, meeting papers, newspaper reports, patents, writings, standards and research reports.
B. Constructing a document reference model based on document reference information, deleting self-guide interference, and generating a learner reference network by linear mapping;
in this embodiment, the "building a document citation model based on document citation information, deleting self-introduction interference, and generating a student citation network by linear mapping" includes the following steps:
s31, constructing a document citation network model, counting the citation times and the other citation times of the document, and then executing the step S32;
and S32, generating a self-introduced document other-introduced network model according to the document other-introduced times mapping, and mapping to generate a student cited network.
Among the above schemes, the ith reference in S31 is in reference group GliteratureBy referencing variablesIs obtained by accumulation, and the calculation formula isThe ith literature in step S31 is in literature group GliteratureBy referencing a variableAnd coefficient of self-inductionProduct of (2)Is obtained by accumulation, and the calculation formula isIf the jth document cites the ith document, thenEqual to 1; if the jth document does not refer to the ith document, thenEqual to 0; if there is at least one scholar identity between the ith and jth documents, and the citation is self-citation, thenEqual to 0; if the ith article and the jth article do not have the same scholars, and the citation is other citations, thenEqual to 1; wherein i is more than or equal to 1, and j is more than or equal to 1; in the step S32, the number of times of the reference is determinedGenerating a self-introduced excluded document tare network model Gliterature-citedAnd then generates the learner reference network G by linear mappingcitation. In this embodiment, as shown in fig. 5, the learner reference network is constructed as follows: gcitation=(V,Ecitation),Gcitation=(V,Ecitation) Is composed of N scholar nodes and Ecitation|=McitationA directed network formed by edges; wherein G iscitationThe representative scholars quote the scholars and the collection of the relationship quoted by the scholars in the network, V represents the scholars group GcitationSet of trainees in (E)citationRepresentative group of scholars GcitationReference relations between the Chinese scholars;representative scholars quote scholars v in networkiAnd scholars vjIn the student, the reference relationship set GcitationChinese scholar viBy other scholars vjThe number of references totals up toIf the scholar viV. a certain scientific research result in scholarsjIf the scientific research result is quoted, the number is marked as 1,is a scholar viAnd scholars vjSum of directed edges, i.e. scholar viIs learnt of scientific achievements vjThe sum of the number of references of (a); if the scholar viHas no learners vjIf the reference is made, marking as 0; wherein i is more than or equal to 1, j is more than or equal to 1,andthe same or different. The student node distribution of the student reference network is the same as that of the student cooperative network, and the difference is that the edges representing the node relationship in the student reference network and the student cooperative network are different.
C. Fusing a student cooperation network and a student reference network to generate a student relation network;
in this embodiment, as shown in fig. 6, the learner relationship network is formed by merging a learner cooperation network and a learner reference network:
G=α·Gcollaboration+β·Gcitation;
wherein α, β are assigned values for weights of a scholars cooperative network and a scholars reference network, α + β is 1, assignments of α and β can be calculated and assigned through a computing terminal or can be set manually, the scholars relation model is G (V, E), wherein,the scholars relationship network is a directed network consisting of | V | ═ N scholars nodes and | E | ═ M edges; wherein G represents the set of scholars and relationships among scholars in the scholars group, V represents the set of scholars in the scholars group G, E represents the relationships among scholars in the scholars group G, EijStudent v in representative student relationship networkiAnd scholars vjIn the context of (a) or (b),is a node viThe strength of the penetration of (A) into (B),is a node viThe strength of (2).
D. And calculating important nodes in the scholars relationship network and clustering and grouping, wherein the result is the selected expert list.
In this embodiment, as shown in fig. 3, the step of calculating important nodes in a learner relationship network includes performing importance ranking on a learner in a learner cooperation network according to an importance ranking method of nodes in a complex network;
preferably, the method for ranking according to the importance of the nodes in the complex network includes: the scholars are sorted by a method of recursively peeling off the network based on node strength, which specifically comprises the following steps:
s11, setting k as an integer, and taking k as 0;
s12, removing all student nodes with the strength not greater than k in the student relationship network, and simultaneously deleting edges connected with the student nodes with the strength not greater than k;
s13, checking the output intensity of the rest student nodes in the network at the moment, judging whether the student nodes with the intensity not greater than k still exist in the rest student nodes, if so, executing a step S12, otherwise, executing a step S14;
s14, forming a kth layer by the scholar nodes with the intensity not greater than k removed in the step S12, endowing the nodes of the kth layer with a node Ks value, namely equal to k, and then executing a step S15;
s15, judging whether the number of the nodes left in the network is 0, if so, executing a step S17, otherwise, executing a step S16
S16, setting k to k +1, and repeatedly executing step S12;
s17, sorting the scholars in a descending order according to the Ks values.
In this embodiment, as shown in fig. 4, the "performing cluster grouping" includes performing community division on a scholar relationship network, where the divided scholar communities are regarded as relatively independent research fields, and the steps are as follows:
s21, taking the scholar relationship network G as an initial network and setting the initial network as a current network;
s22, randomly dividing the nodes in the current network into two communities, and then executing a step S23;
s23, calculating the contribution degree of each node to the modularity degree, calculating the network modularity degree according to the contribution degree, and then executing the step S24;
s24, moving the nodes with lower contribution degree from one community to another community, and then executing the step S25;
s25, recalculating the contribution degree of each node to the modularity and the network modularity, and then executing the step S26;
s26, judging whether the network modularity is increased or not, simultaneously judging whether the network modularity reaches the maximum value or not, if the network modularity is increased, namely the maximum value is not reached, keeping the moving result of the node and returning to the step S24, and if the network modularity is not increased, withdrawing the mobile node, moving a new node with lower contribution degree different from the withdrawn mobile node from one community to another community, and returning to the step S25; if the modularity reaches the maximum value, executing step S27;
s27, recording and storing the network modularity and community structure of the initial network at the moment, and then executing a step S28;
and S28, continuously dividing each community divided in the step S27 as an individual network, and executing the steps S22-S28 for each individual network in a recursive mode until no larger modularity is generated in the initial network, so as to obtain a network community division result.
The contribution degree lambda of each node to modularity degreeiCalculated according to the following formula:
wherein, κr(i)Representing nodes v belonging to a community riThe sum of the edge values representing the relationship with other nodes in the community,is a node viThe strength of the penetration of (A) into (B),is a node viA strength ofr(i)Representing the ratio of the variable values of the nodes in the community r.
The modularity Q is calculated according to the following equation:
wherein m is the sum of the edge values representing the relationship in the scholars relationship network.
In this embodiment, the selection method further includes extracting student information in the database metadata information, and performing a data cleaning step, where in the data cleaning step, the "student name + primary organization + secondary organization" is used as a unique identifier of each student, and an ID number is assigned.
In this embodiment, for a journal paper with keywords containing "high performance fiber" as an example, 10,547 scientific papers and 123,653 scholars are selected first; establishing a student cooperation network model and a student reference network model; fusing to generate a scholars relationship network model; and carrying out node importance arrangement calculation on the student relationship network G obtained by fusion, selecting student nodes with higher Ks values in the student relationship network G to form a new student relationship network, then carrying out community division on the newly formed student relationship network, wherein the division result is a plurality of student communities corresponding to each sub-technical field in the technical field of the student, forming an expert list according to each sub-field, and obtaining 1,075 experts belonging to 16 research fields.
In the above scheme, when the research field included in the specific technical topic is less in branches, if the learner relationship network G is directly clustered and grouped, a large number of expert learners may exist in each sub-field, and the lead learners in the research field cannot be intuitively obtained in the list. In addition, when identifying authoritative experts in a specific technical subject who are in a research hotspot, the calculation process of clustering first and then sorting learners is increased by meaningless calculation amount compared with the process of sorting first and then clustering. Aiming at the problem, the importance of the student nodes in the student relational network is preferentially sorted, a threshold value can be set to determine the student nodes with higher Ks so as to form a new student relational network G ", and then clustering grouping is carried out, so that a more intuitive expert list can be obtained, and more important expert and students in research hotspots of various sub-fields are embodied.
In this embodiment, for example, the conference paper whose keywords contain "artificial intelligence", 11,432 conference papers are selected first, and 25,985 scholars are selected; establishing a student cooperation network model and a student reference network model; fusing to generate a scholars relationship network model; and carrying out community division on the student relationship network G obtained by fusion, wherein the division result is a plurality of student communities corresponding to each sub-technical field under the technical subject to which the student belongs, on the basis, respectively carrying out node importance arrangement calculation on each sub-field student relationship network obtained by division (namely a plurality of new student relationship networks divided by the student relationship network G), obtaining students with higher Ks values in each sub-field student relationship network to form an expert list, and obtaining 2,563 experts belonging to 22 research fields.
In the above scheme, when the research field included in a specific technical subject is numerous and complicated, if a method of directly sorting the importance of nodes in the student relationship network G to obtain an expert list is adopted, the experts in the hot sub-technical field are ranked higher, and the experts in the cold sub-technical field are not selected into the list due to the ranking, which may result in the loss of some experts. Aiming at the problem, the invention preferentially adopts clustering grouping to divide the sub-fields of the student relationship network G, and then carries out importance sequencing, thereby effectively avoiding missing expert students in the cold field.
In this embodiment, the user manually checks the domain expert lists in order.
Although the present invention has been described with reference to a preferred embodiment, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. A method for selecting domain experts based on a citation network and a scientific research cooperation network is characterized by comprising the following steps:
firstly, constructing a student cooperative directed network model based on database metadata information to generate a student cooperative network;
secondly, constructing a document reference model based on document reference information, deleting self-induced interference, and generating a learner reference network by linear mapping;
then, fusing the student cooperation network and the student reference network to generate a student relation network;
finally, calculating important nodes in the scholars relationship network and clustering and grouping, wherein the result is a selected expert list;
wherein the student cooperative network is a directed network.
2. The method of claim 1, wherein the learner cooperation network is a method for selecting domain experts based on a citation network and a scientific research cooperation network, the method comprising: gcollaoration=(V,Ecollaboration),Gcollaoration=(V,Ecollaboration) Is composed of N scholar nodes and Ecollaboration|=McollaborationA directed network of edges, wherein GcollborationRepresents a set of scholars and scholars' cooperative relations in a scholars cooperative network, V represents a scholars group GcollborationSet of trainees in (E)collaborationRepresentative group of scholars GcollaborationThe cooperative relationship among the Chinese scholars;
representative of scholars viAs a collaboration center, scholars viAnd scholars vjThe cooperative relationship of (1); in-scholar partnership collection GcollborationMiddle school student viAs a collaboration center, scholars viThe total number of times of cooperation with other scholars isIf the scholar viAnd scholars vjThere is a common achievement of scientific research, and the scholars viCollaborate as a result of the studyThe central scholar, then scholar viAnd scholars vjThere is a cooperation between them, denoted 1,is a scholar viAnd scholars vjSum of number of edges, i.e. by scholar viAs a collaboration center, scholars viAnd scholars viSum of the number of collaborations of; otherwise, the learner viAnd scholars vjThere is no cooperative relationship between them, and it is marked as 0; wherein i is more than or equal to 1, j is more than or equal to 1,andthe same or different;
the cooperation center scholars can calculate distribution through a computing terminal and can also set manually;
preferably, a first person in charge of scientific and technological projects, a communicator of journal papers, a communicator of conference papers, a first contributor of newspaper reports, a first inventor of patents, a first main edition of writings, a first standard drafter and a first reporter of research reports are selected as cooperation center scholars; if the journal paper does not indicate a communication student, the first student of the journal paper is used as a cooperation center student; if the conference paper does not indicate a communication student, the first student of the periodical paper is used as a cooperation center student;
more preferably, the first person in charge of the scientific research is taken as a scholarer of the cooperation center.
3. The method for field expert selection based on citation network and scientific research cooperation network as claimed in claim 1, wherein said student citation network is: gcitation=(V,Ecitation),Gcitation=(V,Ecitation) Is composed of N scholar nodes and Ecitation|=McitationA directed network formed by edges; wherein G iscitationRepresentatives of the academic societyThe student quotes the scholars and the collection of the relation quoted by the scholars in the network, and V represents the scholars group GcitationSet of trainees in (E)citationRepresentative group of scholars GcitationReference relations between the Chinese scholars;
representative scholars quote scholars v in networkiAnd scholars vjIn the student, the reference relationship set GcitationChinese scholar viBy other scholars vjThe number of references totals up toIf the scholar viV. a certain scientific research result in scholarsjIf the scientific research result is quoted, the number is marked as 1,is a scholar viAnd scholars vjSum of directed edges, i.e. scholar viIs learnt of scientific achievements vjIs the number of citations sum of scholars viHas no learners vjIf the reference is made, marking as 0; wherein i is more than or equal to 1, j is more than or equal to 1,andthe same or different;
the student node distribution of the student reference network is the same as that of the student cooperative network, and the difference is that the edges representing the node relationship in the student reference network and the student cooperative network are different.
4. The method for domain expert selection based on citation network and scientific research cooperation network as claimed in claims 1-3, wherein said student relationship network is formed by combining student cooperation network and student citation networkG is α Gcollaboration+β·Gcitationα and β are weight distribution values of the trainee cooperative network and the trainee reference network, the α + β is 1, and the assignment values of α and β can be calculated and distributed through a computing terminal or can be set manually;
the scholars relationship model G ═ (V, E)
the scholars relationship network is a directed network consisting of | V | ═ N scholars nodes and | E | ═ M edges; wherein G represents the set of scholars and relationships among scholars in the scholars group, V represents the set of scholars in the scholars group G, E represents the relationships among scholars in the scholars group G, EijStudent v in representative student relationship networkiAnd scholars vjIn the context of (a) or (b),is a node viThe strength of the penetration of (A) into (B),is a node viThe strength of (2).
5. The method for field expert selection based on citation network and scientific research cooperative network as claimed in claim 1, wherein said calculating important nodes in student relationship network comprises ranking the importance of students in student cooperative network according to the importance ranking method of nodes in complex network;
preferably, the method for ranking according to the importance of the nodes in the complex network includes: the scholars are sorted by a method of recursively peeling off the network based on node strength, which specifically comprises the following steps:
s11, setting k as an integer, and taking k as 0;
s12, removing all student nodes with the strength not greater than k in the student relationship network, and simultaneously deleting edges connected with the student nodes with the strength not greater than k;
s13, checking the output intensity of the rest student nodes in the network at the moment, judging whether the student nodes with the intensity not greater than k still exist in the rest student nodes, if so, executing a step S12, otherwise, executing a step S14;
s14, forming a kth layer by the scholar nodes with the intensity not greater than k removed in the step S12, endowing the nodes of the kth layer with a node Ks value, namely equal to k, and then executing a step S15;
s15, judging whether the number of the nodes left in the network is 0, if so, executing a step S17, otherwise, executing a step S16
S16, setting k to k +1, and repeatedly executing step S12;
s17, sorting the scholars in a descending order according to the Ks values.
6. The method for field expert selection based on citation network and scientific research cooperation network as claimed in claim 1, wherein said clustering comprises community division of student relationship network, the divided student community is regarded as relatively independent research field, and the steps are as follows:
s21, taking the scholar relationship network G as an initial network and setting the initial network as a current network;
s22, randomly dividing the nodes in the current network into two communities, and then executing a step S23;
s23, calculating the contribution degree of each node to the modularity degree, calculating the network modularity degree according to the contribution degree, and then executing the step S24;
s24, moving the nodes with lower contribution degree from one community to another community, and then executing the step S25;
s25, recalculating the contribution degree of each node to the modularity and the network modularity, and then executing the step S26;
s26, judging whether the network modularity is increased or not, simultaneously judging whether the network modularity reaches the maximum value or not, if the network modularity is increased, namely the maximum value is not reached, keeping the moving result of the node and returning to the step S24, and if the network modularity is not increased, withdrawing the mobile node, moving a new node with lower contribution degree different from the withdrawn mobile node from one community to another community, and returning to the step S25; if the modularity reaches the maximum value, executing step S27;
s27, recording and storing the network modularity and community structure of the initial network at the moment, and then executing a step S28;
and S28, continuously dividing each community divided in the step S27 as an individual network, and executing the steps S22-S28 for each individual network in a recursive mode until no larger modularity is generated in the initial network, so as to obtain a network community division result.
7. The method as claimed in claim 6, wherein the contribution λ of each node to modularity is determined by a domain expert selection method based on a citation network and a scientific research cooperation networkiCalculated according to the following formula:
wherein, κr(i)Representing nodes v belonging to a community riThe sum of the edge values representing the relationship with other nodes in the community,is a node viThe strength of the penetration of (A) into (B),is a node viA strength ofr(i)Representing the ratio of the variable values of the nodes in the community r.
9. The method for selecting the domain experts based on the citation network and the scientific research cooperative network as claimed in claim 1, wherein the step of building a document citation model, deleting self-citation interference and generating a student citation network by linear mapping based on document citation information comprises the following steps:
s31, constructing a document citation network model, counting the citation times and the other citation times of the document, and then executing the step S32;
and S32, generating a self-introduced document other-introduced network model according to the document other-introduced times mapping, and mapping to generate a student cited network.
10. The method for selecting domain experts based on citation network and scientific research cooperation network as claimed in claim 1, wherein the selecting method comprises extracting student information in database metadata information and performing a data cleaning step, wherein the data cleaning step uses a student name + primary organization + secondary organization as a unique identifier of each student and assigns an ID number.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911154794.6A CN111078873B (en) | 2019-11-22 | 2019-11-22 | Domain expert selection method based on citation network and scientific research cooperation network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911154794.6A CN111078873B (en) | 2019-11-22 | 2019-11-22 | Domain expert selection method based on citation network and scientific research cooperation network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111078873A true CN111078873A (en) | 2020-04-28 |
CN111078873B CN111078873B (en) | 2021-02-09 |
Family
ID=70311264
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911154794.6A Active CN111078873B (en) | 2019-11-22 | 2019-11-22 | Domain expert selection method based on citation network and scientific research cooperation network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111078873B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112000811A (en) * | 2020-08-25 | 2020-11-27 | 北京搜狗科技发展有限公司 | Doctor information processing method and device |
CN112100370A (en) * | 2020-08-10 | 2020-12-18 | 淮阴工学院 | Picture examination expert combined recommendation method based on text convolution and similarity algorithm |
CN112732889A (en) * | 2020-12-07 | 2021-04-30 | 东南大学 | Student retrieval method and device based on cooperative network |
CN113269653A (en) * | 2021-06-18 | 2021-08-17 | 北京市科学技术情报研究所 | Social network management method and system based on circled thought |
CN113868407A (en) * | 2021-08-17 | 2021-12-31 | 北京智谱华章科技有限公司 | Evaluation method and device for review recommendation algorithm based on scientific research big data |
CN114328673A (en) * | 2021-12-31 | 2022-04-12 | 杭州师范大学 | Scientific research personnel data processing method based on complex network |
CN115630141A (en) * | 2022-11-11 | 2023-01-20 | 杭州电子科技大学 | Scientific and technological expert retrieval method based on community query and high-dimensional vector retrieval |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160231979A1 (en) * | 2012-03-07 | 2016-08-11 | Salesforce.Com, Inc. | Verification of shared display integrity in a desktop sharing system |
CN109002524A (en) * | 2018-07-13 | 2018-12-14 | 北京市科学技术情报研究所 | A kind of gold reference author's sort method based on paper adduction relationship |
CN109375888A (en) * | 2018-09-07 | 2019-02-22 | 北京奇艺世纪科技有限公司 | A kind of throwing screen method and device |
-
2019
- 2019-11-22 CN CN201911154794.6A patent/CN111078873B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160231979A1 (en) * | 2012-03-07 | 2016-08-11 | Salesforce.Com, Inc. | Verification of shared display integrity in a desktop sharing system |
CN109002524A (en) * | 2018-07-13 | 2018-12-14 | 北京市科学技术情报研究所 | A kind of gold reference author's sort method based on paper adduction relationship |
CN109375888A (en) * | 2018-09-07 | 2019-02-22 | 北京奇艺世纪科技有限公司 | A kind of throwing screen method and device |
Non-Patent Citations (1)
Title |
---|
贺焕振: "基于合著与引用网络的专家知识地图构建研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112100370A (en) * | 2020-08-10 | 2020-12-18 | 淮阴工学院 | Picture examination expert combined recommendation method based on text convolution and similarity algorithm |
CN112100370B (en) * | 2020-08-10 | 2023-07-25 | 淮阴工学院 | Picture-trial expert combination recommendation method based on text volume and similarity algorithm |
CN112000811A (en) * | 2020-08-25 | 2020-11-27 | 北京搜狗科技发展有限公司 | Doctor information processing method and device |
CN112732889A (en) * | 2020-12-07 | 2021-04-30 | 东南大学 | Student retrieval method and device based on cooperative network |
CN113269653A (en) * | 2021-06-18 | 2021-08-17 | 北京市科学技术情报研究所 | Social network management method and system based on circled thought |
CN113269653B (en) * | 2021-06-18 | 2024-03-29 | 北京市科学技术情报研究所 | Social network management method and system based on layering thought |
CN113868407A (en) * | 2021-08-17 | 2021-12-31 | 北京智谱华章科技有限公司 | Evaluation method and device for review recommendation algorithm based on scientific research big data |
CN114328673A (en) * | 2021-12-31 | 2022-04-12 | 杭州师范大学 | Scientific research personnel data processing method based on complex network |
CN114328673B (en) * | 2021-12-31 | 2024-04-16 | 杭州师范大学 | Scientific research personnel data processing method based on complex network |
CN115630141A (en) * | 2022-11-11 | 2023-01-20 | 杭州电子科技大学 | Scientific and technological expert retrieval method based on community query and high-dimensional vector retrieval |
CN115630141B (en) * | 2022-11-11 | 2023-04-25 | 杭州电子科技大学 | Scientific and technological expert retrieval method based on community query and high-dimensional vector retrieval |
Also Published As
Publication number | Publication date |
---|---|
CN111078873B (en) | 2021-02-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111078873B (en) | Domain expert selection method based on citation network and scientific research cooperation network | |
Murata et al. | Link prediction of social networks based on weighted proximity measures | |
Seker | Computerized argument Delphi technique | |
Meijering et al. | Exploring research priorities in landscape architecture: An international Delphi study | |
CN108446886A (en) | Personnel recruitment system and method based on big data | |
Asanbe et al. | Teachers’ performance evaluation in higher educational institution using data mining technique | |
CN109800300A (en) | learning content recommendation method and system | |
CN110990662B (en) | Domain expert selection method based on citation network and scientific research cooperation network | |
CN111078859B (en) | Author recommendation method based on reference times | |
Kashid et al. | A review of mathematical multi-criteria decision models with a case study | |
Li | Research on evaluation method of physical education teaching quality in colleges and universities based on decision tree algorithm | |
Qureshi et al. | Information needs & information seeking behavior of students in Universities of Pakistan | |
Carmona et al. | Subgroup discovery in an e-learning usage study based on Moodle | |
Siswanto et al. | Implementation of decision support system for campus promotion management using fuzzy multiple analytic decision making (FMADM) method (Case study: Universitas multimedia nusantara) | |
Sutrisno et al. | Application of fuzzy multiple criteria decision making (MCDM) in selection of prospective employees | |
Islam et al. | Management decision-making by the analytic hierarchy process: a proposed modification for large-scale problems | |
Wei et al. | Topic detection based on weak tie analysis: A case study of LIS research | |
Nazim et al. | Information searching habits of Internet users: A users’ study of Banaras Hindu University | |
Lewis | How transdisciplinary is design? An analysis using citation networks | |
Procaci et al. | How Do Outstanding Users Differ From Other Users in Q&A Communities? | |
Apriyani et al. | Decision Support System Location of Development Center Using Promethee Method | |
Elayidom et al. | Applying data mining techniques for placement chance prediction | |
Sun | [Retracted] Mathematical Modeling and Simulation of Online Teaching Effect Evaluation Based on Decision Tree Algorithm | |
Huang | Analysis of public physical education teaching and quality evaluation in colleges and universities based on decision tree algorithm | |
Wahyono et al. | Matching User in Online Learning using Artificial Intelligence for Recommendation of Competition |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |