CN113254642A - E-government affair project evaluation expert group recommendation method based on multi-dimensional feature balance - Google Patents
E-government affair project evaluation expert group recommendation method based on multi-dimensional feature balance Download PDFInfo
- Publication number
- CN113254642A CN113254642A CN202110589302.7A CN202110589302A CN113254642A CN 113254642 A CN113254642 A CN 113254642A CN 202110589302 A CN202110589302 A CN 202110589302A CN 113254642 A CN113254642 A CN 113254642A
- Authority
- CN
- China
- Prior art keywords
- expert
- knowledge
- project
- evaluation
- expert group
- 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
- 238000011156 evaluation Methods 0.000 title claims abstract description 67
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000012552 review Methods 0.000 claims description 11
- 238000005065 mining Methods 0.000 claims description 9
- 238000005516 engineering process Methods 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 2
- 230000007774 longterm Effects 0.000 claims description 2
- 238000000605 extraction Methods 0.000 abstract description 12
- 230000002996 emotional effect Effects 0.000 abstract description 10
- 230000001149 cognitive effect Effects 0.000 abstract description 3
- 238000004458 analytical method Methods 0.000 description 10
- 238000005259 measurement Methods 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 6
- 230000008451 emotion Effects 0.000 description 6
- 230000011218 segmentation Effects 0.000 description 6
- 230000006870 function Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000009469 supplementation Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
- G06F16/355—Class or cluster creation or modification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The electronic government affair project has more administrative connotations, so that the knowledge structure of the electronic government affair project evaluation expert group plays an important role in project management. Based on cognitive scientific theory and psychological theory, multiple characteristic differences of expert individuals are the root causes of cognitive differences and emotional polarity differences. The question whether the expert group is more effective or not as a group decision method for balancing the difference is actually established on the composition method for balancing various characteristics of the expert group. Therefore, the expert group recommendation method for characterizing the expert characteristics and realizing the characteristic balance has important basic value for the consistency of the electronic government affair project evaluation and the improvement of the overall level of the project evaluation. The result is used in an expert extraction stage before project evaluation, and can greatly improve the conformity and the multi-dimensional feature balance between an expert group and an evaluated project, thereby improving the result consistency level of the project evaluation.
Description
1. Field of the invention
Computer science and technology (artificial intelligence, knowledge engineering), scientific and technical management and intelligence, knowledge measurement, knowledge management, text processing technology
2. Background of the invention
(1) Science and technology management and intelligence
Scientific research focuses more on factors such as consistency of expert research expertise and evaluated items, expert reading, authority and the like, and aims to achieve peer evaluation as far as possible and solve the problem of cognitive difference caused by complex knowledge boundaries. The research of intelligence focuses on the discovery method of experts with specific knowledge types, and is one of the methods of scientific research.
In the scientific and technical management, the meta-evaluation theory provides a method for understanding the quality deviation of expert evaluation aiming at the problem of expert evaluation. The meta-evaluation indexes mainly comprise objective indexes such as deviation coefficients and variation coefficients constructed based on the scoring data.
(2) Knowledge metrology
Knowledge metrology is a basic theory for expert expertise identification, and knowledge elements are defined in the knowledge metrology as basic units of knowledge metrology. The expert opinions as a knowledge unit are the analogy basis composed of different knowledge elements according to different structures. The scientific literature is actually different knowledge units, and knowledge elements can be obtained in the scientific literature through text mining technology.
On the basis of expert feature definition guided by an evaluation target, the expert opinions accumulated in the process of evaluating the electronic government affair project are used as a data source of main element evaluation, the multi-dimensional features of the experts are depicted by using basic methods of knowledge mining and knowledge measurement, and therefore the problem pertinence and the rationality of the expert group recommendation method based on multi-dimensional feature balance are achieved.
(3) Knowledge management
And (4) taking a project evaluation grading table with time continuity of the electronic government project evaluation expert and an expert opinion short text as data sources, and performing text mining on the expert opinion to complete knowledge acquisition. And evaluating the level of the expert by taking the comprehensive scoring of the expert group and the opinion text of the expert group formed at the same time as an analogy basis. Comprises the following contents:
establishing a project knowledge concept tree: the project knowledge concept tree is actually a hierarchical knowledge representation method with standard concepts for evaluation targets. The project knowledge concept tree is used for supporting hierarchical semantic recognition of expert review opinion entities, and can support SAO structure-based semantic mapping after structural processing is carried out on expert opinion texts based on SAO structure dependency syntax analysis, so that semantic recognition of different concept levels is carried out on entities in the expert opinion texts.
Establishing an electronic government project knowledge ontology: the project knowledge ontology can completely express the project background knowledge and is the semantic specification of knowledge fusion. The concept and the relation between the concepts of the E-government project knowledge ontology come from the relevant standards and regulations of E-government project management.
Third, opinion mining technology: opinion mining is an effective means for knowledge acquisition in expert opinion texts. The difficulty of short text opinion mining is that the expression of opinion elements in short text has a variety of forms or non-explicit modes, which is more difficult to understand in the context of Chinese. Automated short text understanding requires reliance on additional knowledge that can help machines fully mine word-to-word relationships in short text, such as semantic relatedness.
Fourthly, knowledge aggregation technology: from a knowledge management perspective, the knowledge structure of an expert group may be characterized by the aggregation of the knowledge structures of multiple experts in the group. The emotion polarity and evaluation experience of the expert group can be presented through scientific measurement, and then the multi-dimensional features of the expert group can be calculated.
(4) Text processing techniques
Conventional text processing techniques: the method comprises the steps of sorting a project evaluation index scoring table, training a concept word bank of a project by utilizing the standard and management standard of an e-government project, and assisting with a basic word bank formed by synonyms and near-synonyms. The method comprises the steps of utilizing an open source tool Hanlp to perform sentence segmentation, word segmentation, part of speech tagging and meaningless stop words removal on an expert opinion text, utilizing a dependency syntax analysis recognition statement SAO (principal and predicate object) structure to perform division of an opinion sentence with multiple subjects or multiple objects, and refining a complex sentence containing multiple evaluation objects into a plurality of SAO structure simple sentences containing only unique evaluation objects.
(5) Emotion analysis techniques
And judging the emotional tendency of the expert opinion sentences by using a deep learning model, and performing emotional classification comparison by using four classifiers of RNN, LSTM, BIRNN and BILSTM to obtain an optimal opinion set with determined semantics and determined emotional tendency.
3. Summary of the invention
On the basis of expert feature definition guided by an evaluation target, the achievement takes expert opinions accumulated in the evaluation process of an e-government project as main element evaluation content, and can realize the targeted multi-dimensional feature depiction of experts by using basic methods of knowledge mining and knowledge measurement, thereby completing the expert group recommendation based on multi-dimensional feature balance. The specific steps and methods are described as follows:
(1) knowledge supplementation and representation
In order to realize acquisition and classification of knowledge elements in expert opinions, a project knowledge concept tree of a micro-layer is defined to express the semantic hierarchy between project concept knowledge and concepts in a layering manner; introducing a domain knowledge ontology to represent project knowledge in a macro layer, and supporting knowledge expansion and association and knowledge transformation to realize implicit knowledge discovery; thereby supporting the goal of acquisition and classification of the knowledge elements. The knowledge of the concept tree and the ontology comes from national standards and relevant management policies for e-government projects.
(2) Knowledge acquisition based on hierarchical semantic recognition
The expert opinion knowledge acquisition aims at acquiring knowledge elements in expert opinions and provides a basis for expert knowledge structure description and emotion polarity analysis. The strategy is used for knowledge mining and acquisition in a vocabulary layer (word segmentation and part of speech tagging), a syntax layer (named entity recognition and grammar analysis) and a semantic layer (semantic recognition) respectively.
Aiming at the short text features of the opinion to be reviewed, SAO (subject-action-object) extraction is carried out on the expert opinion on the basis of syntactic analysis to achieve acquisition of structured knowledge, then hierarchical semantic recognition is supported by a concept tree, expert opinion evaluation features are mined out, and an emotion classifier is built by utilizing a deep learning model to obtain emotion polarity. The part specifically comprises the following contents:
named entity recognition
The HMM, CRF, BILSTM and BILSTM-CRF named entity recognition methods are integrated through a voting method to obtain a better result in the extraction of the government function entity. The expert review opinion text is taken as data, firstly, an open source tool Hanlp is utilized to preliminarily classify the expert opinion text into sentences, participles, part of speech tagging and stop words are removed, and then the participle characteristics Xi and the part of speech characteristics POSi are fused into named entity tagging to improve the effect of an integrated model. And forming a domain entity word stock while extracting.
Second, intrinsic semantic analysis and structured knowledge acquisition
The dependency syntactic analysis is utilized to split the opinion sentences with multiple subjects or multiple objects in parallel, fine-grained SAO structured knowledge is extracted and used as a basic semantic unit to represent expert opinions, the subjects S and the objects O mainly represent evaluated objects, generally terms, dynamic terms and the like, and subjects or objects formed by the words or phrases are closely related to project concept semantics.
Evaluation characteristic layered semantic recognition
The evaluation opinion after SAO structuring needs evaluation feature layering semantic recognition aiming at the subject and the object to obtain the fine-grained evaluation feature with strong pertinence. Because different experts have differences in concept expression, the concept semantic mapping based on the concept tree is comprehensively realized by adopting an accurate matching method and a fuzzy matching method based on the maximum similarity of texts so as to obtain a better layered semantic recognition effect, and the method comprises the following steps:
1) precise matching method
For SAO structure text SAOiSubject to it SiAnd object OiRespectively with concepts C in concept tree concept set CjPerforming bidirectional maximum matching, and merging the concept sets successfully matched with the two to obtain a set Ri. If R isiNot null, compare RiThe depth of each concept in (1) is the maximum depthThe concept is a semantic recognition result of the sentence; if R isiAnd if the result is null, fuzzy matching is carried out.
2) Fuzzy matching method
In the process of forming the structured opinion text, in order to not damage the integrity of the expert opinions, modified words such as adjectives, adverbs, prepositions, conjunctions and the like are reserved. However, in the fuzzy matching calculated by using the text similarity, the word segmentation is required to be utilized, the modified words are removed through part-of-speech screening to improve the accuracy, and the main word set S is reservedi’＝{si1，si2，...，simAnd Oi’＝{oi1，oi2，...，oinAnd (m and n are the number of words).
Thus, concept C in concept set CjWord segmentation is carried out to obtain cj＝{cj1，cj2，...，cjtAnd (t is the number of words). Similarity Sim (S) based on fine-grained participlesi’，cj)、Sim(Oi’，cj) The calculation process of (2) is shown in the formula (1).
Wherein sim(s)ik，cjr) And calculating the cosine similarity of the word2vec word vector. The result obtained by averaging the similarity in a Cartesian product form through fine-grained word segmentation has higher accuracy. Set of cyclic concepts C, compute Sim (S)i’，cj)、Sim(Oi’，cj) Taking the maximum similarity greater than the threshold, the corresponding concept is the SAOiIdentified concept semantics.
Definition of unstructured knowledge acquisition
Due to the nonstandard problem of language writing in the short text, the syntactic analysis effect of part of opinions is poor, and the extraction of SAO structured knowledge is inevitably wrong, so that the internal semantic relation in the text cannot be recognized. The three main questions presented in the opinion are summarized and the relevant rules are customized to relocate the knowledge as shown in table 1.
TABLE 1 expert opinion non-SAO structured knowledge semantic analysis
(3) Expert multi-dimensional feature selection and calculation
And (4) selecting the expert characteristics to have clear target constraint-electronic government project evaluation knowledge constraint. The expert evaluates the project based on the self knowledge structure, and is essentially a knowledge exchange activity between an evaluation subject and an evaluation object. The electronic government project is used as an information project for government function construction, and has more requirements on the breadth of the expert knowledge structure. Accordingly, with reference to the domain concept tree and the domain knowledge ontology, four constituent elements of the expert knowledge structure are defined for knowledge classification, as shown in table 2.
Table 2 expert knowledge structure definition table
By referring to the basic theory of the meta-evaluation theory and the objective meta-evaluation index, the knowledge level, the evaluation profound, the emotional style and the field expertise are defined as the multi-dimensional characteristic description indexes of the experts. The definition comprehensively considers the knowledge ability of experts and psychological factors influencing the review of the experts and the acquirability of related information. The intrinsic logic of the expert features is explained in table 3.
TABLE 3 expert multidimensional characteristic evaluation index
(ii) knowledge level
The index measure is designed according to the knowledge level definition in table 2 as formula (3-5):
Wherein
m represents the number of the expert groups in the evaluation activity.
LeveltFor the knowledge level index of an expert in the knowledge structure t element, by a weight alphalAnd betalControlling the importance of scoring biases and opinion biases. n represents the number of item elements corresponding to the knowledge component. DjTo score the deviation factor, it is shown that for the item element j, the expert scores xjIs equally divided by all experts
Relative deviation of (d). O isjAn opinion deviation coefficient representing the relative deviation, sk, of the average emotional intensity of the opinions of the individual experts and the average emotional intensity of the entire expert group for the item element j1j、sk2jRespectively representing the opinion sentiment strength of individual experts and expert groups on the project element j, K1、K2The individual expert and expert group opinion numbers belonging to the project element j, respectively.
(ii) evaluation of deep gravities
The more profound the evaluation opinions of the experts, the more the essential knowledge of the information used in the construction of the e-government project can be grasped. Therefore, an expert evaluation deep index and a measuring method thereof based on the semantic hierarchy of the domain concept tree are provided to depict the expert knowledge depth.
Expert review professionalism measures the information described by the opinion from several aspects:
the characteristic vocabulary of the E-government field related in the expert opinions is as long as the opinions, if the vocabulary is large, the opinion content is likely to be rich and profound, and the expert knowledge breadth and depth are large.
The frequency of appearance of some characteristic words is higher, so that the concept related to the characteristic is more definite and the pertinence is stronger.
The level, the distribution path and the concentration (node output degree) of the feature words distributed in the concept tree determine the specific and definite degree of the evaluated feature semantic content, and related knowledge structural elements have pertinence and prominence or the knowledge forms a key point.
Wherein the RFtAnd RFtThe count is the evaluation characteristic set and the evaluation characteristic number thereof belonging to a certain knowledge structure in the opinion respectively, and the rf-countiIs the number of ith rating features in the rating opinions.
Definition 1: the frequency of the evaluation features k belonging to the element t of the knowledge structure in the expert review opinions is shown in formula (6):
Definition 2: the concentration of the evaluation feature c is the ratio of the number of occurrences of hyponyms c-son of the evaluation feature c in the evaluation opinions to the number of hyponym set elements. Therefore, the concentration of the evaluation feature k in the expert opinion is given as shown in equation (7):
definition 3: the evaluation deep is the sum of the deep reflected by the frequency and concentration of all evaluation feature words in the opinion in the domain feature concept tree. The calculation formula (8) is as follows:
setting two deviation weights alphal、βlThe knowledge level of each evaluation of the average expert is integrated to obtain the overall knowledge structure level of the expert, wherein the knowledge level is 0.5; performing profound calculation on all evaluation opinions of an expert through concept tree semantic level identification to obtain an expert evaluation profound value, determining the weight of a child concept node to be 1.2 times of the weight of a corresponding parent concept node according to the upper and lower parent-child relations, and setting two profound measurement weights alphad、βdAre all 0.5. In order to clearly understand the knowledge breadth and the knowledge depth of the same expert, the knowledge structure distribution conditions of the expert and the expert are presented in the same radar map.
Third, emotional style
Expert sentiment is delivered by the opinion evaluation, and the total sentiment intensity REMo of the expert opinions is calculated as shown in formula (9) according to the sentiment style definition in Table 2, wherein K is the total number of all opinions of the items that the expert has reviewed, and sk represents the sentiment intensity of the expert opinion K. The result is that the expert can visually review the emotional style by using a histogram.
Expertise in field
The method adopts an LDA method, takes all the evaluation project titles and abstracts as data sources, and analyzes the content subject of the expert evaluation so as to reflect the project field related to the expert. The method mainly comprises the following steps: 1) cleaning data, removing punctuation marks and numbers, and filtering stop words; removing common software description words to improve the theme representation degree (such as modules, services, platforms, systems and the like) of the LDA on government affairs, project functions and functions; and finishing the construction of the word bag. 2) Determining the number of the topics according to the confusion degree to improve the model effect, and simultaneously obtaining a matrix with the relationships between the topics and words and between the documents and the topics by using an LDA model. 3) Analyzing the matrix data to obtain the distribution condition of words under each topic and the condition of the topic to which each document belongs, finally counting the probability of the topic to which each examined item of each expert belongs, and analyzing the past examined content topic of the expert.
(4) Expert group recommendation method for multi-dimensional feature balance
The expert group recommendation method based on multi-dimensional feature balance is based on the measurement result of multi-dimensional features among expert groups, and aims to realize the evaluation consistency among different expert groups. On the basis of random and fair extraction, the recommendation method not only needs to meet the requirement that the domain expertise of the candidate expert group has better closeness with the content of the project to be examined; and the knowledge breadth and the knowledge depth of the candidate expert group are not lower than the average level of all the expert knowledge structure configuration of the expert database.
(ii) evaluation of domain relevance
The subject probability of the project to be evaluated is obtained through the LDA model, and the subject similarity with the expert field is calculated to obtain the correlation degree between the expert and the project field to be evaluated
The larger the value, the higher the correlation between the expert and the domain of the project to be examined, and the calculation formula (10) shows. Wherein
Representing the project P to be evaluatedaThe subject feature vector of (1) is,
presentation expert EiSubject feature vector of
The calculation formula of the average probability under the tth topic of (1) is shown in (11), M represents expert EiThe number of items reviewed.
Knowledge balanced random extraction
The knowledge level balance among the expert groups can be realized by the knowledge complementation of the experts in each group. For this purpose, first, the knowledge level and the evaluation deep-scale average of each knowledge structure of the expert database are calculated and recorded as
And
as a threshold, where i ∈ { government knowledge, technical knowledge, administrative knowledge, budget knowledge }; secondly, the domain correlation obtained in the last stage is taken as a weight omega to be merged into random extraction; finally, randomly extracting m (m is more than or equal to 3 and less than N) experts from an expert database (N experts) to form a candidate expert group, respectively calculating the knowledge level and the evaluation deep scale average value of each knowledge structure of the candidate expert group, and if the knowledge level and the evaluation deep scale average value exceed the threshold value
And
and keeping the extraction result, otherwise, performing random extraction again.
4. Description of the drawings
FIG. 1: e-government project expert review opinion table: data sources of the result are displayed;
FIG. 2: e-government project field knowledge concept tree schematic diagram: showing the e-government field concept tree (part) created by the result;
FIG. 3: e-government project domain ontology graph: displaying an electronic government project field knowledge ontology created by the result;
FIG. 4: and (3) achievement implementation of a framework diagram: the complete process of realizing the result is shown;
FIG. 5: the expert knowledge level and the evaluation deep characteristic depict a result graph: displaying the feature depiction result of a single expert;
FIG. 6: expert emotion style measurement result graph: the emotional style in the multi-expert opinion is displayed;
FIG. 7: expert field expertise visualization result graph: the evaluation field experience of a single expert is displayed;
FIG. 8: the result comparison graph of the result method and the random extraction method is as follows: a comparison of the results of the two methods is shown.
5. Detailed description of the preferred embodiments
The implementation purpose of the achievement is to realize the E-government project expert group recommendation method based on multi-dimensional feature balance, 214 provincial E-government project expert group review opinions in 2017-2018 of Tianjin city are used as data sources, and 50 experts with long-term evaluation experience in an expert library are used for feature depiction and expert group recommendation experiments based on 3-bit and 5-bit experts. The method is used in the expert extraction stage before project evaluation, and the knowledge measurement result of the extraction result of the expert group is proved to be superior to the method of randomly extracting experts by the method of the achievement. The method can greatly improve the conformity and the multi-dimensional feature balance between the expert group and the evaluated project, thereby improving the consistency level of the result of the project evaluation.
The achievement field ontology is stored in Neo4j based on Cypher language; the experiment is realized in a Windows environment by utilizing a pyhanlp package provided by an open source tool Hanlp, Google open source deep learning framework Tensorflow, a high-level API (application program interface) -keras and the like based on Python language.
Claims (4)
1. A method for recommending an item evaluation expert group in the field of electronic government affairs is characterized by comprising the following steps: s1, defining expert characteristic indexes, and acquiring a plurality of dimensional characteristics of project evaluation experts in the field of electronic government affairs by using a knowledge metering method; and S2, uniformly extracting experts based on the multidimensional characteristics, and improving the relevance between the expert groups and the evaluated items and the consistency of the expert groups on the evaluation of the electronic government field items.
2. The method of claim 1, wherein the method utilizes unstructured text, long-term historical data of expert review opinions; and acquiring knowledge elements in the opinions of the individual experts by using a text-based mining technology, and realizing knowledge acquisition of the opinions of the expert group through knowledge association and fusion.
3. The method according to claim 1 and 2, characterized in that multidimensional characteristic indexes of the electronic government affairs project review experts are created on a knowledge semantic level, and mainly comprise indexes of knowledge level, review depth and the like.
4. The method according to claims 1 and 3, wherein the method comprises an expert group recommendation method based on multi-dimensional feature equalization.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110589302.7A CN113254642A (en) | 2021-05-28 | 2021-05-28 | E-government affair project evaluation expert group recommendation method based on multi-dimensional feature balance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110589302.7A CN113254642A (en) | 2021-05-28 | 2021-05-28 | E-government affair project evaluation expert group recommendation method based on multi-dimensional feature balance |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113254642A true CN113254642A (en) | 2021-08-13 |
Family
ID=77184993
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110589302.7A Pending CN113254642A (en) | 2021-05-28 | 2021-05-28 | E-government affair project evaluation expert group recommendation method based on multi-dimensional feature balance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113254642A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110137726A1 (en) * | 2009-12-03 | 2011-06-09 | Telefonica, S.A. | Recommender system based on expert opinions |
| CN102479202A (en) * | 2010-11-26 | 2012-05-30 | 卓望数码技术（深圳）有限公司 | Recommendation system based on domain expert |
| CN109062961A (en) * | 2018-06-27 | 2018-12-21 | 淮阴工学院 | A kind of expert's combination recommended method of knowledge based map |
| CN110443574A (en) * | 2019-07-25 | 2019-11-12 | 昆明理工大学 | Entry convolutional neural networks evaluation expert's recommended method |
| CN111709604A (en) * | 2020-05-19 | 2020-09-25 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Evaluation method and device for expert review behavior and computer storage medium |
| CN111782797A (en) * | 2020-07-13 | 2020-10-16 | 贵州省科技信息中心 | Automatic matching method for scientific and technological project review experts and storage medium |
-
2021
- 2021-05-28 CN CN202110589302.7A patent/CN113254642A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110137726A1 (en) * | 2009-12-03 | 2011-06-09 | Telefonica, S.A. | Recommender system based on expert opinions |
| CN102479202A (en) * | 2010-11-26 | 2012-05-30 | 卓望数码技术（深圳）有限公司 | Recommendation system based on domain expert |
| CN109062961A (en) * | 2018-06-27 | 2018-12-21 | 淮阴工学院 | A kind of expert's combination recommended method of knowledge based map |
| CN110443574A (en) * | 2019-07-25 | 2019-11-12 | 昆明理工大学 | Entry convolutional neural networks evaluation expert's recommended method |
| CN111709604A (en) * | 2020-05-19 | 2020-09-25 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Evaluation method and device for expert review behavior and computer storage medium |
| CN111782797A (en) * | 2020-07-13 | 2020-10-16 | 贵州省科技信息中心 | Automatic matching method for scientific and technological project review experts and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| 朱柏: "项目评审中的专家组的选择", ***工程理论与实践, no. 03, pages 76 - 80 * |
| 王忠群等: "基于领域专家和商品特征概念树的在线商品评论深刻性度量", 《现代图书情报技术》, pages 17 - 25 * |
| 陈媛等: "科研项目同行评议专家水平的评价研究", 《科学学与科学技术管理》, pages 38 - 42 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lalata et al. | A sentiment analysis model for faculty comment evaluation using ensemble machine learning algorithms | |
| Tighe et al. | Personality Trait Classification of Essays with the Application of Feature Reduction. | |
| Medagoda et al. | Sentiment lexicon construction using SentiWordNet 3.0 | |
| CN113934814B (en) | Automatic scoring method for subjective questions of ancient poems | |
| Rakhsha et al. | Detecting adverse drug reactions from social media based on multichannel convolutional neural networks modified by support vector machine | |
| Budzynska et al. | Argument Mining. | |
| Addepalli et al. | A proposed framework for measuring customer satisfaction and product recommendation for ecommerce | |
| Rabin et al. | Development of the formal model for the presentation of poorly structured and unstructured information | |
| Ezzat et al. | Topicanalyzer: A system for unsupervised multi-label arabic topic categorization | |
| Bouarara | Sentiment analysis using machine learning algorithms and text mining to detect symptoms of mental difficulties over social media | |
| Rizun et al. | Methodology for text classification using manually created corpora-based sentiment dictionary | |
| Gurin | Methods for Automatic Sentiment Detection | |
| Shaikh et al. | An intelligent framework for e-recruitment system based on text categorization and semantic analysis | |
| Widoyono et al. | Sentiment analysis of learning from home during pandemic covid-19 in indonesia | |
| CN113254642A (en) | E-government affair project evaluation expert group recommendation method based on multi-dimensional feature balance | |
| Søyland et al. | Party polarization and parliamentary speech | |
| Bellar et al. | Sentiment Analysis of Tweets on Social Issues Using Machine Learning Approach | |
| Sulayes | Quantitative authorship attribution of users of Mexican drug dealing related online forums | |
| Yu et al. | Text Classification Based on Natural Language Processing and Machine Learning in Multi Label Corpus | |
| Tabaza et al. | An adaptive intelligent framework for assessment & selection process in staffing task | |
| Glass et al. | Codes of Ethics: Extending Classification Techniques with Natural Language Processing | |
| Shekhar | Text Mining and Sentiment Analysis | |
| Pandi et al. | Reputation based online product recommendations | |
| Doti | Natural Language Processing for Survey Analysis | |
| Whalley et al. | Sentiment lexicon construction using sentiwordnet 3.0 |
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 |