Institution recommendation system and method based on scientific and technological service needs

Abstract

The present invention discloses an institution recommendation system and method based on technology service demand, the system includes a data preparation module, a technology service demand analysis module, a technology service institution evaluation module and a technology service platform recommendation module; the data preparation module is used to collect technology service data in each technology service platform; the technology service demand analysis module is used to find a recommended set of technology service institutions that have transactions with neighboring technology service demanders; the technology service institution evaluation module is used to calculate the comprehensive score of technology service institutions; the technology service platform recommendation module is used to select a technology service institution recommendation list and send it to the demander to be recommended. The present invention comprehensively considers the demand content of the technology service demander and the ability level of the technology service institution, recommends a list of institutions of interest and high ability to the demander, and greatly improves the accuracy and effectiveness of platform recommendations.

Description

Technical Field

[0001] The present invention relates to the technical field of science and technology service organization recommendation, and in particular to an organization recommendation system and method based on science and technology service demand. Background Art

[0002] The science and technology service industry has linked science and technology with industry in a two-way manner, greatly promoting the industrialization of knowledge and technology, driving the economic growth of the whole society through R&D and innovation, and driving the transformation and upgrading of industries. With the rapid development of the science and technology service industry, the science and technology service platform has attracted more and more science and technology service institutions to settle in. In the process of continuous aggregation of resources, the science and technology service platform has also generated a large amount of heterogeneous information.

[0003] On the one hand, the increase in technology service agencies has injected fresh and high-quality technology service resources into the platform, enabling the platform to better and faster meet technology service needs, achieve information sharing and effective cooperation, and help technology service agencies and even the technology service industry advance; but on the other hand, it has also increased the difficulty for technology service demanders on the platform to accurately find high-quality and suitable technology service agencies. In the face of massive amounts of information, after technology service demanders upload their needs, traditional search functions are not convenient for them to quickly screen out satisfactory agencies. At this time, the platform needs to realize the automatic matching of technology service agencies with technology service needs, and recommend suitable technology service agencies to technology service demanders.

[0004] At present, the research on science and technology service platforms at home and abroad is still focused on the exploration of platform construction and layout, which is mainly divided into the service model, service process, function and significance, incentive and performance evaluation of science and technology service platforms. Only some studies have described from a theoretical perspective how to carry out personalized recommendations on science and technology service platforms, promote information display and interaction, coordinate resource sharing and release, and improve the accuracy and efficiency of supply and demand matching. For example, "Research on Cloud Service Model of Regional Science and Technology Resource Sharing Platform under Big Data Environment" discusses the framework of using recommendation systems to meet the needs of science and technology services. For potential needs, the platform should actively recommend hot information and information of interest to users through mining and processing of user information; for future needs, the platform integrates and mines user information and national policy information to predict the needs that users may have in the future and make corresponding recommendations. In addition, "Research on Jining Science and Technology Cloud Service Platform" specifically proposes to use latent semantic model algorithms to mine and process user information, and actively recommend hot spots and content that may be of interest to users. Therefore, there is still a lack of systems and methods for conducting personalized institutional recommendations on science and technology service platforms from an experimental perspective.

[0005] In the personalized recommendation of institutions, existing research mainly uses the basic information, behavior logs, scores and other information of institutions and demanders for recommendation research, and also considers the trust between demanders and institutions, the reputation of institutions and other information to improve the recommendation effect. However, the research and analysis of the demand content (demand title and demand description) uploaded by technology service demanders is not sufficient, and there is little comprehensive evaluation of institutions by mining the specific comments of technology service demanders on institutions. Summary of the invention

[0006] The purpose of the present invention is to improve the accuracy and effectiveness of the institution recommendation list by analyzing the demand content uploaded by the technology service demander, the technology service platform rating and the technology service institution evaluation labels, and to provide an institution recommendation system and method based on technology service demand.

[0007] To achieve the above-mentioned purpose, the present invention designs an institution recommendation system based on technology service demand, which is used to send a list of technology service institution recommendations to the demander to be recommended. The system is special in that it includes a data preparation module, a technology service demand analysis module, a technology service institution evaluation module and a technology service platform recommendation module;

[0008] The data preparation module is used to collect technology service data about the demanders to be recommended, other technology service demanders, and technology service agencies that have purchase associations with other technology service demanders in each technology service platform, and pre-process all collected technology service data to obtain valid technology service data;

[0009] The technology service demand analysis module includes a technology service demand model building module and a technology service demand similarity calculation module;

[0010] The technology service demand model construction module is used to receive the demand titles and demand descriptions uploaded to each technology service platform by the demand parties to be recommended and other technology service demand parties sent by the data preparation module, perform Chinese word segmentation on the demand titles and demand descriptions, remove stop words in the word segmentation results, and obtain new demand title texts and new demand description texts; then vectorize each new demand title text and new demand description text composed of terms to obtain new demand title vectors and new demand description vectors; finally, calculate the weight of each term in the new demand title vector and the new demand description vector, and the new demand title vector and the new demand description vector containing the terms and corresponding weights constitute the demand title and demand description VSM model of the demand parties to be recommended and other technology service demand parties, combine the demand title and demand description VSM models, and generate the technology service demand VSM model of the demand parties to be recommended and other technology service demand parties;

[0011] The technology service demand similarity calculation module is used to calculate the similarity of the two dimensions of demand title and demand description respectively according to the technology service demand VSM model of the demander to be recommended and other technology service demanders; then the similarity of the demand title dimension and the similarity of the demand description dimension are weighted and calculated respectively to obtain the demand similarity value of the demander to be recommended and other technology service demanders, and the demand similarity values ​​are sorted from high to low in order of demand similarity values, and the first k nearest technology service demanders with highly similar demands to the demander to be recommended are selected; finally, by comparing the purchase association, a recommended set of m technology service agencies that have transactions with the first k nearest technology service demanders are found;

[0012] The technology service agency evaluation module is used to set the comprehensive scoring weight, weighted calculation platform for the technology service agency's platform score and other technology service demanders' label scores for the technology service agency, to obtain the technology service agency's comprehensive score; it includes a platform score analysis module and an agency evaluation label analysis module;

[0013] The platform rating analysis module is used to find the platform rating given by each technology service platform sent by the data preparation module to the technology service agency that has a purchase association with each other technology service demander;

[0014] The institution evaluation label analysis module is used to find the institution evaluation labels given by each other technology service demander to the technology service institution with which it has a purchase association, sent by the data preparation module, and mark the institution evaluation labels to obtain institution characteristics; then the marked institution characteristics are divided into sentiment tendencies to obtain positive institution characteristics and negative institution characteristics; finally, frequency calculation is performed to calculate each positive institution characteristic score and each negative institution characteristic score respectively, and all positive institution characteristic scores and negative institution characteristic scores are cumulatively added to generate a label score;

[0015] The technology service platform recommendation module is used to receive a set of m technology service agency recommendations sent by the technology service demand analysis module, and receive comprehensive scores of technology service agencies sent by the technology service agency evaluation module; rearrange the m technology service agency recommendation sets in descending order according to the comprehensive scores of the technology service agencies, and select the top n agencies as the final technology service agency recommendation list and send them to the recommended demander.

[0016] The present invention also designs a method for recommending institutions based on scientific and technological service needs, comprising the following steps:

[0017] Step 1), the data preparation module collects technology service data from other technology service platforms, and pre-processes the technology service data to obtain effective technology service data;

[0018] Step 2), the technology service demand model construction module performs Chinese word segmentation on the demand titles and demand descriptions uploaded to each technology service platform by the recommended demander and other technology service demanders, removes stop words in the word segmentation results, and obtains new demand title texts and new demand description texts; constructs new demand title vectors and new demand description vectors, extracts terms in the new demand title vectors and new demand description vectors, calculates the weights corresponding to the terms, and establishes demand title and demand description VSM models; combines the demand title and demand description VSM models to generate a technology service demand VSM model;

[0019] Step 3), based on the VSM model of technology service demand, the technology service demand similarity calculation module takes weights for the two dimensions of demand title and demand description, respectively, and performs weighted calculation to obtain the demand similarity value between the to-be-recommended demander and other technology service demanders, and selects the top k nearest technology service demanders whose demands are highly similar to those of the to-be-recommended demander according to the demand similarity value; finally, by comparing the purchase association, find the set of m technology service institutions that have transactions with the top k nearest technology service demanders;

[0020] Step 4), based on the technology service platform transaction information, the platform rating analysis module searches for the platform rating of each technology service platform for the technology service institution;

[0021] Step 5) Based on the transaction information of the technology service platform, the institution evaluation label analysis module searches for the institution evaluation labels of other technology service demanders for technology service institutions, screens out positive and negative institutional characteristics, and then obtains each positive institutional characteristic score and each negative institutional characteristic score according to the frequency calculation formula, and the cumulative addition can generate the label score of the technology service institution;

[0022] Step 6) The science and technology service agency evaluation module sets the comprehensive score weight to ensure that the platform score weight is lower than the label score weight, and weights the platform score and label score to obtain the comprehensive score of the science and technology service agency;

[0023] Step 7) The technology service platform recommendation module rearranges the set of m technology service agencies in descending order of their comprehensive scores, selects the first n agencies as the final technology service agency recommendation list and sends it to the recommended demander.

[0024] The advantages of the present invention are:

[0025] 1. The present invention takes the demand for scientific and technological services as the core and vectorizes the unstructured demand text uploaded by the party in need of scientific and technological services. This personalized representation method helps to accurately meet the demand for scientific and technological services and improve the accuracy of recommendations;

[0026] 2. The present invention constructs a comprehensive score for science and technology service agencies based on the platform score and the agency evaluation label of science and technology service agencies, measures the capability level of science and technology service agencies, and helps recommend agencies with high capability levels to science and technology service demanders to improve the effectiveness of the recommendation;

[0027] 3. Based on the institutional evaluation labels of science and technology service institutions, positive institutional characteristics and negative institutional characteristics are distinguished according to different sentiment tendencies. Then, according to different scoring methods, the positive institutional characteristic scores and negative institutional characteristic scores are cumulatively added to generate the label score of the science and technology service institutions, thus realizing the quantification of institutional evaluation labels;

[0028] The system and method for recommending institutions based on technology service demand of the present invention comprehensively considers the demand content of technology service demanders and the capability level of technology service institutions, and recommends a list of institutions of interest and high capability to the demanders, thereby greatly improving the accuracy and effectiveness of platform recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a structural block diagram of the institution recommendation system based on scientific and technological service needs of the present invention;

[0030] Figure 2 A flowchart of the method for recommending institutions based on the needs of scientific and technological services of the present invention;

[0031] In the figure: data preparation module 1, science and technology service demand analysis module 2, science and technology service organization evaluation module 3, science and technology service platform recommendation module 4;

[0032] The data preparation module 1 includes: a data acquisition module 1-1 and a data preprocessing module 1-2;

[0033] The technology service demand analysis module 2 includes: a technology service demand model building module 2-1, and a technology service demand similarity calculation module 2-2;

[0034] The science and technology service agency evaluation module 3 includes: a platform rating analysis module 3-1 and an agency evaluation label analysis module 3-2. DETAILED DESCRIPTION

[0035] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments, but the embodiments should not be construed as limiting the present invention.

[0036] like Figure 1 The institution recommendation system based on technology service demand is used to send a list of technology service institution recommendations to the recommended demanders, which includes a data preparation module 1, a technology service demand analysis module 2, a technology service institution evaluation module 3 and a technology service platform recommendation module 4.

[0037] The data preparation module 1 is used to collect technology service data about the recommended demanders, other technology service demanders, and technology service agencies that have purchasing relationships with other technology service demanders in each technology service platform, and pre-process all collected technology service data to obtain valid technology service data.

[0038] The technology service demand analysis module 2 includes a technology service demand model construction module 2-1 and a technology service demand similarity calculation module 2-2.

[0039] The technology service demand model construction module 2-1 is used to receive the demand titles and demand descriptions uploaded to each technology service platform by the recommended demand party and other technology service demand parties sent by the data preparation module 1, perform Chinese word segmentation on the demand titles and demand descriptions respectively, remove stop words in the word segmentation results, and obtain new demand title texts and new demand description texts; then vectorize each new demand title text and new demand description text composed of terms to obtain new demand title vectors and new demand description vectors; finally, calculate the weight of each term in the new demand title vector and the new demand description vector, and the new demand title vector and the new demand description vector containing the terms and corresponding weights constitute the demand title and demand description VSM model of the recommended demand party and other technology service demand parties, combine the demand title and demand description VSM models, and generate the technology service demand VSM model of the recommended demand party and other technology service demand parties.

[0040] The technology service demand similarity calculation module 2-2 is used to calculate the similarity of the two dimensions of demand title and demand description according to the technology service demand VSM model of the recommended demander and other technology service demanders; then take weights for the similarity of the demand title dimension and the similarity of the demand description dimension for weighted calculation to obtain the demand similarity value between the recommended demander and other technology service demanders, sort them in order from high to low according to the demand similarity value, and select the top k neighboring technology service demanders that are highly similar to the demand of the recommended demander; finally, compare with the purchase association to find the recommended set of m technology service agencies that have transactions with the top k neighboring technology service demanders.

[0041] The technology service agency evaluation module 3 is used to set the comprehensive scoring weight, weighted calculation platform for the technology service agency's platform score and other technology service demanders' label scores for the technology service agency, to obtain a comprehensive score for the technology service agency; it includes a platform score analysis module 3-1 and an agency evaluation label analysis module 3-2.

[0042] The platform rating analysis module 3 - 1 is used to find the platform rating given by each technology service platform sent by the data preparation module 1 to the technology service agency that has a purchase association with each other technology service demander.

[0043] The institution evaluation label analysis module 3-2 is used to find the institution evaluation labels given by each other technology service demander to the technology service institution with which it has a purchasing relationship, sent by the data preparation module 1, and mark the institution evaluation labels to obtain the institution characteristics; then the marked institution characteristics are artificially divided into sentiment tendencies to obtain positive institution characteristics and negative institution characteristics; finally, frequency calculation is performed to calculate each positive institution characteristic score and each negative institution characteristic score respectively, and all positive institution characteristic scores and negative institution characteristic scores are cumulatively added up to generate a label score.

[0044] The technology service platform recommendation module 4 is used to receive the m technology service agency recommendation sets sent by the technology service demand analysis module 2, and receive the technology service agency comprehensive scores sent by the technology service agency evaluation module 3; rearrange the m technology service agency recommendation sets in descending order according to the comprehensive scores of the technology service agencies, and select the first n agencies as the final technology service agency recommendation list and send them to the recommended demand party.

[0045] The present invention comprehensively considers the demand content of technology service demanders and the capability level of technology service institutions, and recommends a list of institutions of interest and high capability to the demanders, thereby greatly improving the accuracy and effectiveness of platform recommendations.

[0046] Preferably, the data preparation module 1 includes a data collection module 1-1 and a data preprocessing module 1-2. The data collection module 1-1 uses the Octopus collector to collect technology service data from other technology service platforms; the data preprocessing module 1-2 uses data reduction, data cleaning, and data transformation methods to preprocess the technology service data collected by the data preparation module 1 to extract effective technology service data.

[0047] Preferably, the technology service data includes the demand title and demand description actively uploaded to the platform by the recommended demander, and the technology service platform transaction information of which the transaction process has been completely completed; the technology service platform transaction information includes the transaction information of other technology service demanders, the transaction information of the technology service institutions corresponding to other technology service demanders, and the transaction amount between the two; the transaction information of other technology service demanders includes the name of other technology service demanders, the demand title and demand description actively uploaded to each platform, and the institution evaluation label information made by other technology service demanders to the technology service institutions; the transaction information of the technology service institutions includes the name of the technology service institutions, the institution rating and institution evaluation label made by the platform to the technology service institutions;

[0048] The valid data of technology services include the demand title and demand description actively uploaded to the platform by the recommended demander, as well as the valid technology service platform transaction information whose transaction process has been completely completed; the valid technology service platform transaction information refers to the transaction information excluding the transaction amount less than 100 yuan, the names of other technology service demanders are meaningless, the names of technology service agencies are meaningless, and the agency evaluation labels are invalid (containing only two evaluation labels: "good" and "good comments").

[0049] Preferably, the technology service demand model building module 2-1 uses the Jieba library in Python language to perform Chinese word segmentation on the demand titles and demand descriptions uploaded to each technology service platform by the recommended demander and other technology service demanders. Jieba library is an excellent third-party Chinese word segmentation library in Python language, supporting three word segmentation modes: precise mode, full mode and search engine mode.

[0050] For example, the technology service demand model construction module 2-1 introduces some self-built stop word lists, such as "demand", "need", "company", "enterprise", and "institution", based on the "Harbin Institute of Technology stop word list.txt", "Baidu stop word list.txt", and "Sichuan University Machine Intelligence Laboratory stop word list.txt", and removes the stop words in the word segmentation results to obtain new demand title text and new demand description text.

[0051] Preferably, the new demand title text and the new demand description text are used as input to obtain a VSM model of the technology service demand. Specifically, the technology service demand model construction module 2-1 vectorizes the new demand title text and the new demand description text in the form of 0 and 1 to obtain a new demand title vector and a new demand description vector. Based on the new demand title vector and the new demand description vector, the weights of the terms contained therein are calculated, and the terms and corresponding weights are expressed according to the following formula, that is, a VSM model of the demand title and demand description is formed. The VSM model is:

[0052] d={(t1,w1),(t2,w2),…,(t n ,w n )}

[0053] in,

[0054] d is the new demand title vector or new demand description vector,

[0055] t n For terms in the new requirement title text or new requirement description text,

[0056] w n The weight corresponding to the term in the new requirement title text or the new requirement description text.

[0057] Since the language of the demand title text is generally very concise, the weights corresponding to the terms in the new demand title text are expressed by the word frequency (the number of times the term appears in the new demand title text). Compared with the demand title text, the demand description text is generally longer and more detailed, and belongs to the long text, so the weights corresponding to the terms in the new demand description text are calculated by the classic weight calculation method, that is, the TF-IDF weight calculation formula.

[0058]

[0059] in,

[0060] w i is the weight corresponding to the i-th term in the new requirement description text,

[0061] TF i is the frequency of the i-th term appearing in the new requirement description text,

[0062] IDF i is the inverse document frequency of the i-th term in the new requirement description text,

[0063] The reverse document frequency means that if the new demand description text containing the i-th term is less, that is, n i The smaller the IDF i The larger the value is, the better the ability of the i-th term to distinguish categories is.

[0064] n i is the number of times the i-th term appears in the new requirement description text,

[0065] N is the total number of terms that appear in the new requirement description text,

[0066] l is the total number of new demand description texts, i.e., the total number of technology service demanders.

[0067] df i is the total number of texts containing the i-th term.

[0068] Combine the demand title VSM model and the demand description VSM model to generate the technology service demand VSM model.

[0069] Preferably, based on the VSM model of technology service demand, the technology service demand similarity calculation module 2-2 calculates the similarity between the to-be-recommended demander and other technology service demanders in the two dimensions of demand title and demand description by using the cosine value calculation formula; the similarity in the demand title dimension is weighted as 0.5, and the similarity in the demand description dimension is weighted as 0.5. The weighted calculation is performed with the demand title and the demand description each taking a weight of 0.5 to obtain the similarity value between the to-be-recommended demander and other technology service demanders in terms of technology services.

[0070] Specifically, sort the demand similarity values ​​from high to low, and select the first k nearest technology service demanders that are highly similar to the demand of the recommended demander; finally, compare the purchase associations and find the set of m technology service agencies (m≤k) that have transactions with the first k nearest technology service demanders. Based on the transaction information of the technology service platform, the names of the supply and demand parties that have completely completed the transaction process can be obtained, thereby forming a purchase association relationship between the supply and demand parties on the platform. This relationship is a many-to-many relationship, that is, one demander may correspond to multiple suppliers, and one supplier may correspond to multiple demanders.

[0071] The present invention takes the demand for scientific and technological services as the core and vectorizes the unstructured demand text uploaded by the party in need of scientific and technological services. This personalized representation method helps to accurately meet the demand for scientific and technological services and improve the accuracy of recommendations.

[0072] In addition, the present invention also constructs a comprehensive score for science and technology service agencies based on the two dimensions of the platform score of the science and technology service agency and the agency evaluation label of the science and technology service agency, measures the capability level of the science and technology service agency, and helps to recommend agencies with high capability levels to science and technology service demanders to improve the effectiveness of the recommendation.

[0073] Based on the transaction information of the technology service platform, the platform's rating for each institution can be found. Generally, the platform will evaluate an institution from multiple dimensions, such as consultation response, delivery quality, user evaluation and after-sales service. This rating is a reference factor for the demand side to choose an institution.

[0074] In order to more truly reflect the attitudes and expectations of the demanders towards various scientific and technological service agencies, the present invention also selects the institutional evaluation labels of scientific and technological service agencies for analysis. First, a labeling team is established to determine the labeling standards to fully reflect the unique characteristics of the scientific and technological service industry reflected by the institutional evaluation labels while maximizing the retention of the original institutional evaluation labels. After all the labels are unified, the labels are checked repeatedly to ensure that the labeling results are refined and meet the standards. The labeling results are screened by frequency analysis: the number of occurrences of each institutional feature in the institutional evaluation label is calculated, and institutional features with an occurrence frequency less than a certain threshold are eliminated, and finally multiple institutional features are obtained.

[0075] Based on the institutional evaluation labels of science and technology service agencies, positive institutional characteristics and negative institutional characteristics are distinguished according to different emotional tendencies, and then the positive institutional characteristic scores and negative institutional characteristic scores are cumulatively added according to different scoring methods to generate a label score for the science and technology service agency.

[0076] Preferably, the organization evaluation label analysis module 3-2 divides the sentiment tendency by manually combing the organization characteristics, such as on-time delivery is a positive organization characteristic, and construction delay is a negative organization characteristic, and uses a frequency calculation formula to calculate each positive organization characteristic score and each negative organization characteristic score. The frequency calculation formula is as follows:

[0077]

[0078]

[0079] in,

[0080] S p The score of each positive institutional feature extracted from the institutional evaluation label. The maximum score of each positive institutional feature is 1 point.

[0081] S n The score of each negative institutional feature extracted from the institutional evaluation label. The minimum score of each negative institutional feature is -1.

[0082] F f is the number of times the institution feature appears in the institution evaluation label,

[0083] F 总 The total number of technology service demanders who have given evaluation labels.

[0084] For example, there are 6 evaluations from demand parties in the institution evaluation label, including good (appeared once), very good (appeared twice), very good (appeared three times), and relatively poor (appeared twice). Based on the above evaluation labels, the two institution features of "excellent" and "poor" are extracted. Now, the positive institution feature appears 1+2+3=6 times, and the negative institution feature appears 2 times. f is the number of times this organizational feature appears, F 总 That is 6 (6 demand parties made the evaluation).

[0085] In the above formula, the higher the frequency of positive institutional characteristics, the better, and the lower the frequency of negative institutional characteristics, the better. All positive institutional characteristic scores and negative institutional characteristic scores are cumulatively added together to generate the label score of the science and technology service agency.

[0086] The platform score and the label score are weighted to obtain the comprehensive score of the technology service agency. Considering that the platform score is likely to be higher than the label score, the present invention ensures that the platform score weight is lower than the label score weight when setting the comprehensive score weight of the technology service agency, and finally determines that the weight values ​​of the platform score and the label score are 0.4 and 0.6 respectively.

[0087] The present invention provides a method for recommending an institution based on scientific and technological service needs, comprising the following steps:

[0088] Step 1), the data preparation module 1 collects technology service data from other technology service platforms, and pre-processes the technology service data to obtain effective technology service data.

[0089] Step 2), the technology service demand model construction module 2-1 performs Chinese word segmentation on the demand titles and demand descriptions uploaded to each technology service platform by the recommended demander and other technology service demanders, removes stop words in the word segmentation results, and obtains new demand title text and new demand description text; constructs new demand title vectors and new demand description vectors, extracts terms in the new demand title vectors and new demand description vectors, calculates the weights corresponding to the terms, and establishes demand title and demand description VSM models; combines the demand title and demand description VSM models to generate a technology service demand VSM model.

[0090] Step 3), based on the VSM model of technology service demand, the technology service demand similarity calculation module 2-2 takes weights for the two dimensions of demand title and demand description and performs weighted calculation to obtain the demand similarity value between the recommended demander and other technology service demanders, and selects the top k nearest technology service demanders that are highly similar to the demander to be recommended according to the demand similarity value; finally, by comparing the purchase association, find a recommended set of m technology service agencies that have transactions with the top k nearest technology service demanders.

[0091] Step 4): Based on the technology service platform transaction information, the platform rating analysis module 3-1 searches for the platform ratings of each technology service platform for the technology service institution.

[0092] Step 5) Based on the transaction information of the technology service platform, the institution evaluation label analysis module 3-2 searches for the institution evaluation labels of other technology service demanders for technology service institutions, screens out positive and negative institutional characteristics, and then obtains each positive institutional characteristic score and each negative institutional characteristic score based on the frequency calculation formula. All positive and negative institutional characteristic scores are cumulatively added to generate the label score of the technology service institution.

[0093] Step 6), the science and technology service agency evaluation module 3 calculates the platform score and label score by weighted calculation to ensure that the platform score weight is lower than the label score weight, and obtains the comprehensive score of the science and technology service agency.

[0094] Step 7), the technology service platform recommendation module 4 rearranges the m technology service agency recommendation sets in descending order of the comprehensive scores of the technology service agencies, selects the first n agencies as the final technology service agency recommendation list and sends it to the recommended demander.

[0095] The technical solution of the present invention is described below with reference to embodiments.

[0096] Assume that the demander D1 has uploaded the technology service demand but has not yet matched the technology service agency for the transaction. Then, for the demander D1, according to step 1), the data is collected and pre-processed, as shown in Table 1.

[0097] Table 1

[0098]

[0099] According to step 2), the technology service demand uploaded by D1 is segmented into Chinese words, and stop words are removed to obtain the new demand title text and the new demand description text. The weights of the terms in the new demand title text and the new demand description text are calculated based on the word frequency and TFIDF. The terms and their corresponding weights are combined to form the technology service demand title and demand description VSM model. The VSM model of the demand title and the VSM model of the demand description are then combined to form the technology service demand VSM model.

[0100] According to step 3), the similarity between the recommended demander and other technology service demanders in the two dimensions of demand title and demand description is calculated respectively, and a weighted calculation is performed with a weight of 0.5 for the demand title and demand description respectively, to obtain the similarity value between D1 and other technology service demanders in technology services.

[0101] Sort the similarity values ​​between D1 and other demanders from high to low, and take the first 10 (let k = 10) neighboring technology service demanders to form a list of neighboring technology service demanders, as shown in Table 2. Assuming that neighboring technology service demanders ND1 and ND3 have a purchase association with the same institution, and neighboring technology service demanders ND8 and ND9 have a purchase association with the same institution, then according to the purchase association between these 10 neighboring technology service demanders and institutions, m = 8 institutions can be obtained, forming a recommended set of technology service institutions as shown in Table 2.

[0102] Table 2

[0103]

[0104]

[0105] According to the 8 agency names in the recommended set of science and technology service agencies, according to steps 4) and 5), the platform rating and evaluation label information obtained are shown in Table 3.

[0106] Table 3

[0107]

[0108]

[0109] For the above-mentioned institution evaluation labels, the frequency calculation formula is used to obtain the label score. For example, for science and technology service institution I1, the institution evaluation label is “smooth communication (22)”, where “smooth communication” is a positive institution feature and “(22)” is the number of times “smooth communication” appears in the institution evaluation label.

[0110] According to step 6), the platform score and label score are weighted as 0.4 and 0.6 respectively, and the weighted calculation is performed to obtain the comprehensive score of the science and technology service agency.

[0111] According to step 7), the eight institutions in the recommended set of science and technology service institutions are reordered according to their comprehensive scores, and the first five (n=5) institutions {I8, I6, I2, I1, I4} form a recommended list of science and technology service institutions. The recommended list of science and technology service institutions {I8, I6, I2, I1, I4} is pushed to the science and technology service demander D1 according to the preset notification method.

[0112] The present invention is based on a technology service platform, and utilizes the demands actively uploaded by the demanders to be recommended, the demands actively uploaded by the demanders of other technology service platforms and the agency evaluation labels given by them, the purchase transaction relationships spontaneously stored and recorded by the platform, and the rating information given by the platform to the agencies, to generate a recommendation list of technology service agencies for the demanders to be recommended. For the demanders to be recommended, the agencies ranked at the top in the recommended agency list are of high quality, which is conducive to their acquisition of better service resources to support their own growth and development; for technology service agencies, their probability of being displayed and selected is increased, which is conducive to their own development and growth; for the platform, the resources it recommends are of high quality, and the supply and demand sides are quickly matched, which is conducive to its own long-term development.

[0113] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field.

[0114] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0115] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0116] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0117] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0118] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. An institution recommendation system based on scientific and technological service needs, characterized by: It includes a data preparation module (1), a science and technology service demand analysis module (2), a science and technology service agency evaluation module (3) and a science and technology service platform recommendation module (4); The data preparation module (1) is used to collect technology service data on the technology service demanders to be recommended, other technology service demanders, and technology service agencies that have purchase associations with other technology service demanders in each technology service platform, and pre-process all collected technology service data to obtain valid technology service data; The technology service demand analysis module (2) comprises a technology service demand model construction module (2-1) and a technology service demand similarity calculation module (2-2); The technology service demand model construction module (2-1) is used to receive the demand titles and demand descriptions uploaded to each technology service platform by the to-be-recommended demand party and other technology service demand parties respectively sent by the data preparation module (1), perform Chinese word segmentation on the demand titles and demand descriptions respectively, remove stop words in the word segmentation results, and obtain new demand title texts and new demand description texts; then vectorize each new demand title text and new demand description text composed of terms to obtain new demand title vectors and new demand description vectors; finally, calculate the weight of each term in the new demand title vector and the new demand description vector, and the new demand title vector and the new demand description vector containing the terms and corresponding weights constitute the VSM model of the demand titles and demand descriptions of the to-be-recommended demand party and other technology service demand parties, combine the VSM models of the demand titles and demand descriptions, and generate the technology service demand VSM model of the to-be-recommended demand party and other technology service demand parties; The technology service demand similarity calculation module (2-2) is used to calculate the similarity of the two dimensions of demand title and demand description according to the technology service demand VSM model of the demander to be recommended and other technology service demanders; then, weight the similarity of the demand title dimension and the similarity of the demand description dimension for weighted calculation, and obtain the demand similarity value between the demander to be recommended and other technology service demanders, sort them in descending order according to the demand similarity value, and select the first k nearest technology service demanders whose demands are highly similar to those of the demander to be recommended; finally, compare the purchase association to find the recommended set of m technology service agencies that have transactions with the first k nearest technology service demanders; The technology service agency evaluation module (3) is used to set the comprehensive scoring weight, and weightedly calculate the platform score of the technology service agency and the label score of the technology service agency by other technology service demanders to obtain the comprehensive score of the technology service agency; it includes a platform score analysis module (3-1) and an agency evaluation label analysis module (3-2); The platform rating analysis module (3-1) is used to find the platform rating given by each technology service platform sent by the data preparation module (1) to the technology service agency that has a purchase association with each other technology service demander; The institution evaluation label analysis module (3-2) is used to find the institution evaluation label given by each other technology service demander to the technology service institution with which the data preparation module (1) has a purchase association, and to label the institution evaluation label to obtain the institution characteristics; then to classify the labeled institution characteristics by sentiment tendency to obtain positive institution characteristics and negative institution characteristics; finally to perform frequency calculation, to calculate each positive institution characteristic score and each negative institution characteristic score respectively, to add up all the positive institution characteristic scores and negative institution characteristic scores, and to generate a label score; The technology service platform recommendation module (4) is used to receive the m technology service agency recommendation sets sent by the technology service demand analysis module (2), and receive the technology service agency comprehensive scores sent by the technology service agency evaluation module (3); rearrange the m technology service agency recommendation sets in descending order of the comprehensive scores of the technology service agencies, and select the first n agencies as the final technology service agency recommendation list and send it to the recommended demander.

2. The institution recommendation system based on technology service demand according to claim 1, characterized in that: The technology service demand model construction module (2-1) vectorizes the new demand title text and the new demand description text in the form of 0 and 1 to obtain a new demand title vector and a new demand description vector; the VSM model is expressed as: d={(t1,w1),(t2,w2),…,(t n ,w n )} in, d is the new demand title vector or new demand description vector, t n For terms in the new requirement title text or new requirement description text, w n The weight corresponding to the term in the new requirement title text or the new requirement description text.

3. The institution recommendation system based on technology service demand according to claim 2 is characterized by: The weights of the terms in the new demand title text are expressed by the word frequency, i.e., the number of times the terms appear in the new demand title text; the weights of the terms in the new demand description text are calculated by the TF-IDF weight calculation formula, i.e. in, w i is the weight corresponding to the i-th term in the new requirement description text, TF i is the frequency of the i-th term appearing in the new requirement description text, IDF i is the inverse document frequency of the i-th term in the new requirement description text, n i is the number of times the i-th term appears in the new requirement description text, N is the total number of terms that appear in the new requirement description text, l is the total number of new demand description texts, i.e., the total number of technology service demanders. df i is the total number of texts containing the i-th term.

4. The institution recommendation system based on technology service demand according to claim 3 is characterized by: The institution evaluation label analysis module (3-2) calculates the positive institution characteristic score and the negative institution characteristic score through a frequency calculation formula, and the frequency calculation formula is: in, S p The score of each positive institutional feature extracted from the institutional evaluation label. The maximum score of each positive institutional feature is 1 point. S n The score of each negative institutional feature extracted from the institutional evaluation label. The minimum score of each negative institutional feature is -1. F f is the number of times the institution feature appears in the institution evaluation label, F 总 The total number of technology service demanders who have given evaluation labels.

5. The institution recommendation system based on technology service demand according to claim 1 is characterized by: The technology service demand model building module (2-1) uses the Jieba library in the Python language to perform Chinese word segmentation on the demand titles and demand descriptions uploaded to each technology service platform by the recommended demander and other technology service demanders.

6. The institution recommendation system based on technology service demand according to claim 5 is characterized by: The technology service demand similarity calculation module (2-2) calculates the similarity between the recommended demander and other technology service demanders in the two dimensions of demand title and demand description through the cosine value calculation formula; the similarity in the demand title dimension is weighted as 0.5, and the similarity in the demand description dimension is weighted as 0.

5.

7. The institution recommendation system based on technology service demand according to claim 1 is characterized by: The data preparation module (1) comprises a data collection module and a data preprocessing module; the data collection module uses an Octopus collector to collect technology service data from other technology service platforms; the data preprocessing module uses data reduction, data cleaning, and data transformation methods to preprocess the technology service data collected by the data preparation module (1) to extract effective technology service data.

8. The institution recommendation system based on technology service demand according to claim 1 is characterized by: The technology service data includes the demand title and demand description actively uploaded to the platform by the recommended demander, and the technology service platform transaction information of which the transaction process has been completely completed; the technology service platform transaction information includes the transaction information of other technology service demanders, the transaction information of the technology service institutions corresponding to other technology service demanders, and the transaction amount between the two; the transaction information of other technology service demanders includes the name of other technology service demanders, the demand title and demand description actively uploaded to each platform, and the institution evaluation label information made by other technology service demanders to the technology service institutions; the transaction information of the technology service institutions includes the name of the technology service institutions, the institution rating and institution evaluation label made by the platform to the technology service institutions; The valid data of technology services include the demand title and demand description actively uploaded to the platform by the recommended demander, as well as the valid technology service platform transaction information for which the transaction process has been completely completed; the valid technology service platform transaction information refers to the transaction information excluding the transaction amount less than 100 yuan, the names of other technology service demanders are meaningless, the names of technology service institutions are meaningless, and the institution evaluation labels are invalid.

9. A method for recommending institutions based on scientific and technological service needs, comprising the following steps: Step 1), the data preparation module (1) collects technology service data from other technology service platforms, and pre-processes the technology service data to obtain effective technology service data; Step 2), the technology service demand model construction module (2-1) performs Chinese word segmentation on the demand titles and demand descriptions uploaded to each technology service platform by the recommended demander and other technology service demanders, removes stop words in the word segmentation results, and obtains new demand title texts and new demand description texts; constructs new demand title vectors and new demand description vectors, extracts terms in the new demand title vectors and new demand description vectors, calculates the weights corresponding to the terms, and establishes VSM models of demand titles and demand descriptions; combines the VSM models of demand titles and demand descriptions to generate a VSM model of technology service demand; Step 3), based on the VSM model of technology service demand, the technology service demand similarity calculation module (2-2) takes weights for the two dimensions of demand title and demand description and performs weighted calculation to obtain the demand similarity value between the to-be-recommended demander and other technology service demanders, and selects the first k nearest technology service demanders whose demands are highly similar to those of the to-be-recommended demander according to the demand similarity value; finally, by comparing the purchase association, find a recommended set of m technology service agencies that have transactions with the first k nearest technology service demanders; Step 4), based on the technology service platform transaction information, the platform rating analysis module (3-1) searches for the platform rating of each technology service platform for the technology service institution; Step 5), based on the transaction information of the science and technology service platform, the institution evaluation label analysis module (3-2) searches for the institution evaluation labels of other science and technology service demanders for the science and technology service institution, screens out the positive and negative institution characteristics, and then obtains each positive institution characteristic score and each negative institution characteristic score according to the frequency calculation formula, and adds up all the positive institution characteristic scores and negative institution characteristic scores to generate the label score of the science and technology service institution; Step 6), the science and technology service agency evaluation module (3) calculates the platform score and the label score by weighted calculation, ensuring that the platform score weight is lower than the label score weight, and obtains the comprehensive score of the science and technology service agency; Step 7), the technology service platform recommendation module (4) rearranges the m technology service agency recommendation set in descending order of the comprehensive scores of the technology service agencies, selects the first n agencies as the final technology service agency recommendation list and sends it to the recommended demander.

Images