Achievement transformation data screening method and device, electronic equipment and storage medium

Abstract

This disclosure provides a method, apparatus, electronic device, and storage medium for screening technology transfer data, relating to the field of computer technology. The method includes: inputting the text of the technology to be evaluated into a corresponding knowledge graph for feature extraction to obtain a first feature and a second feature associated with the first feature; determining a corresponding trained evaluation model based on the first and second features; and determining an evaluation result indicating the potential for technology transfer of the text based on the first feature, the second feature, and the trained evaluation model. This improves the efficiency of technology transfer data screening and enhances the accuracy of data screening based on evaluation results.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to a method, apparatus, electronic device and storage medium for screening data related to technology transfer. Background Technology

[0002] The technology transfer data screening is mainly used to filter information with the potential for technology transfer from a massive amount of technical information, and obtain the screening results. The screened results can be recommended to the corresponding users. The screening process mainly involves the evaluation of information.

[0003] In related technologies, the evaluation of information is mainly conducted through the analysis and assessment of experts' subjective knowledge structures. On the one hand, there are few all-round experts, and on the other hand, experts are often quite subjective in their reviews. This leads to some scientific and technological achievements having the technical environment and value for transformation but failing to pass the review, while other scientific and technological achievements lacking the technical environment or economic significance for transformation are approved. The transformed technological achievements have no value or valuable technological achievements are ignored, resulting in low efficiency in the analysis of scientific and technological achievements transformation. Summary of the Invention

[0004] This disclosure provides a method, apparatus, electronic device, and storage medium for screening data related to technology transfer.

[0005] In a first aspect, this disclosure provides a method for screening technology transfer data. The method includes: inputting the technology text to be evaluated into a corresponding knowledge graph for feature extraction, obtaining a first feature and N second features associated with the first feature, wherein N is an integer greater than 0; determining the corresponding trained evaluation model based on the first feature and the second features; and determining the evaluation result that the technology text to be evaluated has the potential for technology transfer based on the first feature, the second features and the trained evaluation model.

[0006] In some embodiments, when N is an integer greater than 1, determining the corresponding trained evaluation model based on the first feature and the second feature includes: determining the corresponding trained first-class evaluation model based on the first feature and each second feature; determining the corresponding trained second-class evaluation model based on the first feature and any two second features; and so on, determining the corresponding trained M-class evaluation model based on the first feature and any M second features, where M is an integer greater than 1 and less than or equal to N.

[0007] In some embodiments, determining the assessment result of the potential for commercialization of the technical text to be assessed based on the first feature, the second feature, and the trained assessment model includes: inputting the first feature and each second feature into the corresponding trained first-type assessment model to obtain N first-type prediction results, and determining a first assessment value based on the N first-type prediction results; inputting the first feature and any two second features into the corresponding trained second-type assessment model to obtain C. 2 N Two second-class prediction results, and according to C 2 N The second-class prediction result is used to determine the second evaluation value; and so on, the first feature and any M second features are input into the corresponding trained M-class evaluation model to obtain C. N M The prediction result for the Mth class, and based on C N M Based on the Mth prediction result, determine the Mth evaluation value; based on the first evaluation value, the second evaluation value, and the Mth evaluation value, determine the evaluation result of whether the technical text to be evaluated has the potential for commercialization.

[0008] In some embodiments, the method further includes: determining the technology category corresponding to the text to be evaluated; and determining the knowledge graph corresponding to the technology text to be evaluated based on the technology category.

[0009] In some embodiments, the method further includes: acquiring sample text data of different technology categories; and constructing a knowledge graph of the technology category based on the sample text data.

[0010] In some embodiments, constructing a knowledge graph of a technology category based on sample text data includes: identifying sample entities in the sample text data; obtaining associated data related to the sample entities in the sample text data; preprocessing the associated data to obtain a set of reference sample entities associated with the sample entities, wherein the set of reference sample entities includes at least one reference associated entity; determining reference sample entities among the reference associated entities based on the correlation between the reference associated entities and the sample entities, and a preset filtering dimension; and constructing a knowledge graph based on the sample entities and the reference sample entities.

[0011] In some embodiments, the method further includes: obtaining a training dataset, wherein the training dataset includes a first sample feature and N second sample feature sets associated with the first sample feature, the second sample feature sets including multiple second sample features; training the evaluation model based on the first sample feature and the second sample feature sets to obtain a trained evaluation model.

[0012] In some embodiments, training the evaluation model based on the first sample features and the second sample feature set to obtain a trained evaluation model includes: training the evaluation model based on the first sample features and the second sample features in each of the second sample feature sets, and the evaluation results of the first type of samples corresponding to the first sample features and the second sample features in each of the second sample feature sets, to obtain a trained first-type evaluation model corresponding to the first sample features and each of the second sample feature sets; training the evaluation model based on the first sample features and the second sample features in any two of the second sample feature sets, and the evaluation results of the second type of samples corresponding to the first sample features and the second sample features in any two of the second sample feature sets, to obtain a trained second-type evaluation model corresponding to the first sample features and any two of the second sample feature sets; and so on, training the evaluation model based on the first sample features and the second sample features in any M of the second sample feature sets, and the evaluation results of the Mth type of samples corresponding to the first sample features and the second sample features in any M of the second sample feature sets, to obtain a trained Mth-type evaluation model corresponding to the first sample features and any M of the second sample feature sets.

[0013] Secondly, this disclosure provides a data screening device for technology transfer, the device comprising: a feature extraction unit, used to input the technical text to be evaluated into a corresponding knowledge graph for feature extraction, obtaining a first feature and N second features associated with the first feature, wherein N is an integer greater than 0; a model determination unit, used to determine the corresponding trained evaluation model based on the first feature and the second features; and a model prediction unit, used to determine the evaluation result that the technical text to be evaluated has the possibility of technology transfer based on the first feature, the second features and the trained evaluation model.

[0014] Thirdly, this disclosure provides an electronic device, including: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the results transformation data screening method disclosed in the embodiments of this disclosure.

[0015] Fourthly, this disclosure provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the results transformation data screening method disclosed in the embodiments of this disclosure.

[0016] Fifthly, this disclosure provides a computer program product, including a computer program that, when executed by a processor, implements the results transformation data screening method disclosed in the embodiments of this disclosure.

[0017] The technical solutions provided by the embodiments of this disclosure have at least the following beneficial effects:

[0018] Input the technical text to be evaluated into the corresponding knowledge graph for feature extraction to obtain the first feature and the second feature associated with the first feature; determine the corresponding trained evaluation model according to the first feature and the second feature; determine the evaluation result of the possibility of achievement transformation of the technical text to be evaluated according to the first feature, the second feature and the trained evaluation model. Based on this, the efficiency of achievement transformation data screening can be improved, and the accuracy of data screening can be improved when screening data based on the evaluation result. Brief Description of the Drawings

[0019] The drawings herein are incorporated into the specification and form a part of this specification, showing embodiments consistent with the present disclosure, and are used together with the specification to explain the principles of the present disclosure.

[0020] Figure 1 It is a flowchart of a method for screening achievement transformation data provided by an embodiment of the present disclosure;

[0021] Figure 2 It is a schematic diagram of a device for screening achievement transformation data provided by an embodiment of the present disclosure;

[0022] Figure 3 It is a schematic diagram of a basic framework of a knowledge graph provided by an embodiment of the present disclosure;

[0023] Figure 4 It is a schematic diagram of another basic framework of a knowledge graph provided by an embodiment of the present disclosure;

[0024] Figure 5 It is a flowchart of a method for constructing a knowledge graph provided by an embodiment of the present disclosure;

[0025] Figure 6 It is a flowchart of a method for training an evaluation model provided by an embodiment of the present disclosure;

[0026] Figure 7 It is a structural diagram of a device for screening achievement transformation data provided by an embodiment of the present disclosure;

[0027] Figure 8 It is a structural diagram of another device for screening achievement transformation data provided by an embodiment of the present disclosure;

[0028] Figure 9 It is a structural diagram of yet another device for screening achievement transformation data provided by an embodiment of the present disclosure;

[0029] Figure 10 It is a structural diagram of yet another device for screening achievement transformation data provided by an embodiment of the present disclosure;

[0030] Figure 11 It is a structural diagram of an electronic device provided by an embodiment of the present disclosure.

[0031] The accompanying drawings have illustrated specific embodiments of this disclosure, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this disclosure to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0032] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0033] The technical solutions of this disclosure and how they solve the aforementioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of this disclosure will now be described with reference to the accompanying drawings.

[0034] The technology transfer data screening is mainly used to filter information with the potential for technology transfer from a massive amount of technical information, and the results can be recommended to the corresponding users.

[0035] The screening process mainly involves evaluating relevant information, and the evaluation of information in related technologies is primarily conducted through the subjective knowledge structure of experts. However, there are few all-around experts, and experts are often quite subjective in their reviews. Some scientific and technological achievements have the technological environment and value for transformation but fail to pass the review, while others lack the technological environment or economic significance for transformation but are still approved. This results in technological achievements having no value or having valuable achievements being overlooked, leading not only to low efficiency in the analysis of technology transfer but also to significant economic losses.

[0036] Based on this, the embodiments of the present disclosure provide an intelligent analysis solution, which can improve the efficiency and accuracy of analysis and evaluation. Since the main analysis and evaluation process can be obtained by extracting according to the analysis and evaluation process of the brain for things and events: extracting the key technical information in things and events, comparing the key technical information with the current technical information to obtain the similarity, and evaluating the event through the similarity. For example, for the analysis and evaluation of 5G short-wave communication technology, by extracting the keywords related to short-wave communication and the associated keywords related to short-wave communication in the report, and then comparing the extracted keywords and associated keywords with the technical information in the stored database, the similarity of the keywords related to short-wave communication and the similarity of the associated keywords related to short-wave communication are obtained. By setting a similarity threshold to sort the similarity, the sorting results of the keywords and associated keywords are obtained. According to the sorting results, the frequency and proportion of the above keywords appearing in the previous technical information are determined. Based on this result, it is determined whether this technical information has technical value, realizing the intelligent evaluation of technical information. It can solve the efficiency problem of only manual evaluation and improve the speed of data screening.

[0037] The above-mentioned intelligent technology information evaluation method, which is also the achievement transformation screening method, mainly obtains the advancement of the technical information in the relevant technology based on the similarity of the technical information keywords in the technical information in the knowledge database of the relevant technology, obtains the possibility of achievement transformation of the technical information through the advancement, and screens this data.

[0038] In some embodiments, based on the problems in the relevant technology, the embodiments of the present disclosure provide a method for screening achievement transformation data, including: inputting the technical text to be evaluated into the corresponding knowledge graph for feature extraction to obtain the first feature and the second feature associated with the first feature; determining the corresponding trained evaluation model according to the first feature and the second feature; determining the evaluation result of the possibility of achievement transformation of the technical text to be evaluated according to the first feature, the second feature and the trained evaluation model. Based on this, the efficiency of screening achievement transformation data can be improved, and the accuracy of data screening can be improved when screening data based on the evaluation result.

[0039] First, in combination with Figure 1 , an exemplary description of the method for screening achievement transformation data provided by the embodiments of the present disclosure will be given.

[0040] Figure 1 It is a flowchart of a method for screening achievement transformation data provided by the embodiments of the present disclosure.

[0041] It should be noted that the technology transfer data filtering method provided in this embodiment can be executed by a technology transfer data filtering device, which can be implemented by software and / or hardware. This technology transfer data filtering device can be an electronic device, or can be configured in an electronic device to enable the electronic device to perform the technology transfer data filtering function. The electronic device in this embodiment can be any device with computing capabilities.

[0042] In some embodiments, such as Figure 2 As shown in the figure, this disclosure provides a data filtering device for technology transfer, including: a technology transfer input module, which provides an interface for users to input technology transfer results into a background database; a background database, which stores the research results input by users and generates directory information; a retrieval module, which receives retrieval requests, generates related terms based on the retrieval requests, and finds the corresponding research results based on the directory information corresponding to the related terms; a technology transfer quality assessment module, which sorts the research results in the background database according to the order of publication and assesses the quality of all research results. When the score of a research result reaches a scoring threshold, the research result is sent to an information push module; an information push module, which receives the research results sent by the research quality assessment module and sends them to an enterprise intelligent display terminal; and the enterprise intelligent display terminal displays the research results sent by the technology quality assessment module. Therefore, the technology transfer data filtering device provided by this disclosure can help improve the efficiency of technology transfer.

[0043] like Figure 1 As shown, the data screening method for this achievement transformation includes:

[0044] S101, input the technical text to be evaluated into the corresponding knowledge graph for feature extraction, and obtain the first feature and the second feature associated with the first feature.

[0045] In this embodiment of the disclosure, the text to be evaluated can be a scientific and technological achievement provided by the user, such as a paper or patent.

[0046] In some embodiments, the technical text to be evaluated is obtained from a background database, wherein the technical text to be evaluated is part or all of the research results provided by the user, and the research results provided by the user are stored in the background database.

[0047] In some embodiments, after determining the text to be evaluated, the technology category corresponding to the text to be evaluated is determined; based on the technology category, the knowledge graph corresponding to the technology text to be evaluated is determined.

[0048] It is understood that in this embodiment of the disclosure, different knowledge graphs are set for texts to be evaluated belonging to different technology categories. After determining the text to be evaluated, the knowledge graph corresponding to the text to be evaluated can be determined according to the technology category corresponding to the text. Based on this, the text to be evaluated is input into the corresponding knowledge graph for feature extraction to obtain a first feature and a second feature associated with the first feature.

[0049] In this embodiment of the disclosure, after obtaining the technical text to be evaluated, the technical text to be evaluated can be input into the corresponding knowledge graph for feature extraction to obtain a first feature and a second feature associated with the first feature.

[0050] In this process, the text of the technology to be evaluated is input into the corresponding knowledge graph for feature extraction, which can obtain at least one first feature and N second features associated with the at least one first feature, where N is an integer greater than 0.

[0051] In some embodiments, the knowledge graph is a ternary structure.

[0052] like Figure 3 As shown, this disclosure provides a basic framework for a knowledge graph.

[0053] like Figure 4 As shown, this disclosure provides another basic framework for knowledge graphs.

[0054] The basic framework of a knowledge graph includes a central first entity interval and multiple second entity intervals associated with the first entity interval. The basic framework also includes attribute definitions for the first and second entity intervals, as well as the attribute relationships between them. The relationship between scientific and technological achievements and technical problems is defined as "the problem to be solved," while "scientific and technological achievements" and "technical problems" are defined as attributes of the first and second entity intervals.

[0055] A knowledge graph can be represented by G = (E, R), where E and R represent the sets of entities and relations of various types in the knowledge graph, respectively. A knowledge graph consists of many RDF triples, represented as T =<Vh,r,vt> Where Vh represents the head entity, vt represents the tail entity, and r represents the relationship between the head entity and the tail entity.

[0056] The construction of knowledge graphs involves the processes of knowledge extraction, knowledge fusion, knowledge processing, and knowledge reasoning from source data.

[0057] The attribute relationship between the first and second entity intervals is defined by adding the relationship between the two related entity intervals on the connecting line of the basic framework. For example, the relationship between "scientific and technological achievements" and "technical methods" can be defined as "inclusion" or "existence"; the relationship between "technical methods" and "technical problems" can be defined as "solution". Alternatively, the attribute relationship between the first and second entity intervals can be determined first, and then the entity data can be filled into the basic framework. The descriptive information and basic characteristics of the achievements include basic information, achievement level, market analysis, transaction information, and other information. Basic information includes: achievement name, brief introduction, research and publication time, application field, key technologies / technical indicators, and patents / papers produced by the achievement; achievement level includes: the technical level of the achievement, innovation points, maturity / stage of the achievement, and application status; market analysis includes: the practical objects of the achievement, application prospects, economic value analysis, market risk analysis, and construction and production conditions; transaction information includes: contact information and transaction methods; other information includes: team, member introductions, affiliated projects, and award information.

[0058] Specifically, for knowledge graphs, the first entity interval is defined as the core interval, and the second entity interval is defined as the expanded interval. The expanded interval is based on the core interval; that is, the entity data filled into the expanded interval is related to and expanded from the entity data filled into the core interval. It can be understood that the entity data filled into the core interval is a larger set, and the entity data filled into the expanded interval are elements within that larger set. Following this approach, the knowledge graph can be understood as expanding using a "general-specific" method to achieve the expansion of entity data and its association with other entity data.

[0059] For feature extraction in knowledge graphs, the node2vec method (i.e., vectorization of network nodes) can be used to train a graph embedding representation vector for each node. This involves sampling the node sequence through random walks and then combining this with the word2vec method to learn a representation for each node in the sequence. In addition to graph embedding representations, each entity class also has corresponding attribute features.

[0060] In this embodiment of the disclosure, the technical text to be evaluated is input into the corresponding knowledge graph for feature extraction to obtain a first feature and a second feature associated with the first feature.

[0061] The first feature can be an entity feature in the first entity interval, and the second feature can be an attribute feature in the second entity interval.

[0062] In this embodiment of the disclosure, the second feature is associated with the first feature, and the first feature and the second feature have an association relationship, which can be an attribute relationship.

[0063] For example, the first feature is X = {researchers, standards, research awards, research projects, papers, patents}, and the second features associated with the first feature are: researchers = {name, department, gender, position, education, age, professional title}, research projects = {project name, department, person in charge, project category, project nature, project source, start time, completion time, approved amount}, and so on.

[0064] S102, Based on the first feature and the second feature, determine the corresponding trained evaluation model.

[0065] In this embodiment of the disclosure, after obtaining the first feature and the second feature, the corresponding trained evaluation model can be determined based on the first feature and the second feature.

[0066] In some embodiments, different first features correspond to different trained evaluation models.

[0067] In some embodiments, for the same first feature, different second features correspond to different trained evaluation models.

[0068] In some embodiments, for the same first feature, different combinations of second features correspond to different trained evaluation models.

[0069] For example, for the same first feature, different combinations of two second features correspond to different trained evaluation models.

[0070] For example, for the same first feature, different combinations of three second features correspond to different trained evaluation models.

[0071] In this embodiment of the disclosure, the corresponding trained evaluation model can be determined based on the first feature and the second feature.

[0072] In some embodiments, the evaluation model is a multi-class support vector machine model employing a radial basis kernel function.

[0073] In this embodiment of the disclosure, the multi-class support vector machine is a type of generalized linear classifier that performs multi-class classification of data using a supervised learning method, and its decision boundary is the maximum margin hyperplane obtained by solving the learning samples.

[0074] S103, based on the first feature, the second feature, and the trained evaluation model, determine the evaluation result of whether the technical text to be evaluated has the potential for commercialization.

[0075] In this embodiment of the disclosure, given the first feature, the second feature, and the trained evaluation model, the evaluation result of whether the technical text to be evaluated has the potential for commercialization can be determined based on the first feature, the second feature, and the trained evaluation model.

[0076] In some embodiments, determining the evaluation result of the potential for commercialization of the technical text to be evaluated based on the first feature, the second feature, and the trained evaluation model includes: mapping the first feature and the second feature to the hyperplane corresponding to the correlation between the first feature and the second feature to obtain the first feature vector and the second feature vector; and inputting the first feature vector and the second feature vector into the trained evaluation model to determine the evaluation result of the potential for commercialization of the technical text to be evaluated.

[0077] For example, the first feature is X = {researchers, standards, research awards, research projects, papers, patents}, and the second feature associated with the first feature is, for example, researchers = {name, department, gender, position, education, age, professional title}, research projects = {project name, department, person in charge, project category, project nature, project source, start time, completion time, approved amount}, and so on.

[0078] Given the triple (h, r, t), TransH first projects the first feature h and the second feature t along the normal wr onto the hyperplane corresponding to the relation r, obtaining the first eigenvector hr and the second eigenvector tr:

[0079] in,

[0080] The triple (h,r,t) satisfies hr+r≈tr.

[0081] Then, the first feature vector hr and the second feature vector tr are input into the trained evaluation model. The output of the trained evaluation model can be the evaluation result of the potential for commercialization of the technology text to be evaluated.

[0082] In this embodiment of the disclosure, features are extracted from the text of the technology to be evaluated through a knowledge graph, and then the evaluation results are obtained using a trained evaluation model. This can improve the efficiency of data screening for the transformation of results. Furthermore, after the evaluation results are determined, it can be determined whether to screen the text of the technology to be evaluated based on the evaluation results. Thus, the accuracy of data screening can be improved when data screening is performed based on the evaluation results.

[0083] In some embodiments, S102, determining the corresponding trained evaluation model based on the first feature and the second feature includes: determining the corresponding trained first-class evaluation model based on the first feature and each second feature; determining the corresponding trained second-class evaluation model based on the first feature and any two second features; and so on, determining the corresponding trained M-th-class evaluation model based on the first feature and any M second features, where M is an integer greater than 1 and less than or equal to N.

[0084] In this embodiment of the disclosure, a first-class evaluation model is determined based on the first feature and each second feature, and a second-class evaluation model is determined based on the first feature and any two second features; and so on, a third-class evaluation model is determined based on the first feature and any M second features, where M is an integer greater than 1 and less than or equal to N.

[0085] For example, given a first feature and two second features, i.e. N = 2, assuming M = N = 2, the corresponding trained first-class evaluation model can be determined based on the first feature and each second feature, and the corresponding trained second-class evaluation model can be determined based on the first feature and the two second features.

[0086] For example, given a first feature and four second features, i.e. N = 4, and assuming M = 3, the corresponding trained first-class evaluation model can be determined based on the first feature and each second feature, the corresponding trained second-class evaluation model can be determined based on the first feature and any two second features, and the corresponding trained third-class evaluation model can be determined based on the first feature and any three second features.

[0087] For example, given two first features, four second features associated with the first first feature, and four second features associated with the second first feature, assuming M is 3, a corresponding trained first-class evaluation model can be determined based on the first first feature and each second feature associated with the first first feature; a corresponding trained first-class evaluation model can be determined based on the second first feature and each second feature associated with the second first feature; a corresponding trained second-class evaluation model can be determined based on the first first feature and any two second features associated with the first first feature; a corresponding trained second-class evaluation model can be determined based on the second first feature and any two second features associated with the second first feature; a corresponding trained third-class evaluation model can be determined based on the first first feature and any three second features associated with the first first feature; and a corresponding trained third-class evaluation model can be determined based on the second first feature and any three second features associated with the second first feature.

[0088] In some embodiments, S103, determining the evaluation result of the potential for commercialization of the technical text to be evaluated based on the first feature, the second feature, and the trained evaluation model includes: inputting the first feature and each second feature into the corresponding trained first-type evaluation model to obtain N first-type prediction results, and determining a first evaluation value based on the N first-type prediction results; inputting the first feature and any two second features into the corresponding trained second-type evaluation model to obtain... The second type of prediction result, and based on The second-class prediction result is used to determine the second evaluation value; similarly, the first feature and any M second features are input into the corresponding trained M-class evaluation model to obtain... The prediction result of the Mth class, and based on Based on the Mth prediction result, determine the Mth evaluation value; based on the first evaluation value, the second evaluation value, and the Mth evaluation value, determine the evaluation result of whether the technical text to be evaluated has the potential for commercialization.

[0089] In this embodiment, the first feature and each second feature are input into the corresponding trained first-class evaluation model to obtain N first-class prediction results. Based on the N first-class prediction results, a first evaluation value is determined. Then, the first feature and any two second features are input into the corresponding trained second-class evaluation model to obtain... The second type of prediction result, and based on The second-class prediction result is used to determine the second evaluation value; similarly, the first feature and any M second features are input into the corresponding trained M-class evaluation model to obtain... The prediction result of the Mth class, and based on Based on the Mth prediction result, determine the Mth evaluation value; based on the first evaluation value, the second evaluation value, and the Mth evaluation value, determine the evaluation result of whether the technical text to be evaluated has the potential for commercialization.

[0090] In some embodiments, a first evaluation value is determined based on N first-class prediction results. The first evaluation value may be the maximum value among the N first-class prediction results, or the minimum value among the N first-class prediction results, or the average value among the N first-class prediction results. This disclosure does not impose specific limitations on this.

[0091] In some embodiments, the evaluation result of the technical text to be evaluated is determined based on the first evaluation value, the second evaluation value, and the Mth evaluation value. The evaluation result may be the maximum value among the first evaluation value, the second evaluation value, and the Mth evaluation value, or the minimum value among the first evaluation value, the second evaluation value, and the Mth evaluation value, or the average value among the first evaluation value, the second evaluation value, and the Mth evaluation value. This disclosure does not impose specific limitations on this.

[0092] In some embodiments, the evaluation result of the technical text to be evaluated is determined based on the first evaluation value, the second evaluation value, and the Mth evaluation value. The first evaluation value, the second evaluation value, and the Mth evaluation value can be sorted and then filtered to determine the evaluation result. For example, the median of the sorted values ​​can be determined as the evaluation result, or the value sorted to a specific position can be determined as the evaluation result. This disclosure does not impose specific limitations on this.

[0093] In some embodiments, a first evaluation value is determined based on N first-type prediction results. This determination can be made through a voting decision.

[0094] For example, the Vth prediction result of the first class is ClassV, 1≤V≤2sqrt(M).

[0095] The voting method is as follows: Class(keyword) = max(F1, F2, ..., FK);

[0096] Where Fk = Fk+1 while ClassV = k, 1≤k≤K, 1≤V≤2sqrt(M).

[0097] For example, given a first feature and two second features, i.e. N = 2, assuming M = N = 2, the first feature and each second feature are input into the corresponding trained first-class evaluation model to obtain two first-class prediction results. Based on the two first-class prediction results, the first evaluation value is determined.

[0098] Furthermore, the first feature and two second features are input into the corresponding trained second-class evaluation model to obtain a second-class prediction result, and the second evaluation value is determined based on the second-class prediction result.

[0099] In some embodiments, the second evaluation value is equal to the second type of prediction result.

[0100] After determining the first evaluation value and the second evaluation value, the evaluation result can be determined based on the first evaluation value and the second evaluation value.

[0101] In some embodiments, the first and second evaluation values ​​can be sorted and then filtered to determine the evaluation result. For example, the median of the sorted values ​​can be determined as the evaluation result, or the value sorted to a specific position can be determined as the evaluation result. This disclosure does not impose specific limitations on this.

[0102] In some embodiments, such as Figure 5 As shown in the embodiments of this disclosure, a method for constructing a knowledge graph is provided, including:

[0103] S501, Obtain sample text data for different technology categories.

[0104] S502, construct a knowledge graph of the technology category based on the sample text data.

[0105] In this embodiment of the disclosure, sample text data of the corresponding technology category is obtained according to different technology categories, and then a knowledge graph of the corresponding technology category is constructed based on the sample text data of the technology category.

[0106] In some embodiments, S502, constructing a knowledge graph of the technology category to which the sample text data belongs includes: determining sample entities in the sample text data; obtaining associated data related to the sample entities in the sample text data; preprocessing the associated data to obtain a set of reference sample entities associated with the sample entities, wherein the set of reference sample entities includes at least one reference associated entity; determining reference sample entities in the reference associated entities based on the correlation between the reference associated entities and the sample entities, and a preset filtering dimension; and constructing a knowledge graph based on the sample entities and the reference sample entities.

[0107] In this embodiment of the disclosure, after obtaining the sample text data, the sample entities in the sample text data can be determined, and the associated data related to the sample entities in the sample text data can be further obtained. The associated data is preprocessed to obtain a set of reference sample entities associated with the sample entities.

[0108] The process involves preprocessing the associated data to obtain a set of data to be screened, and then plotting a scatter plot based on the discrete data in this set. Based on the distribution characteristics of each discrete point in the scatter plot, a fitted curve connecting multiple discrete points and its goodness of fit are determined. If the goodness of fit is less than a preset threshold, a probability distribution model is constructed based on the distance from each discrete point to the fitted curve. This model represents the distribution of the distance from each discrete point to the fitted curve. Based on the mean and standard error of the probability distribution model, and the obtained significance level, a confidence interval for the distance values ​​from all discrete points to the fitted curve is determined. Discrete data corresponding to points whose distance values ​​fall outside the confidence interval are then removed from the set of data to be screened. The set of data remaining after removing this discrete data is designated as a reference set of sample entities associated with the sample entities. This approach allows for the fitting of a curve based on the distribution trend of the discrete data in the set of data to be screened. Using this fitted curve as a benchmark, outliers can be removed, improving the accuracy and thoroughness of outlier removal.

[0109] An initial conversion data filtering network is constructed based on a classification network. The network parameters are updated using unknown technical document data to obtain a conversion data filtering network with stronger generalization ability. The trained conversion data filtering network filters technical document data, which can avoid the mass attack of malicious data and prevent malicious, irrelevant or missing data from being stored, thereby interfering with the accuracy of user search results.

[0110] In the practical example disclosed herein, after determining the set of reference sample entities, the set of reference sample entities includes at least one reference related entity. Based on the correlation between the reference related entity and the sample entity, and the preset screening dimensions, the reference sample entity in the reference related entity is determined. Based on the sample entity and the reference sample entity, a knowledge graph is constructed.

[0111] In some embodiments, such as Figure 6 As shown in the embodiments of this disclosure, a method for evaluating the training of a model is provided, including:

[0112] S601, Obtain the training dataset, wherein the training dataset includes a first sample feature and N second sample feature sets associated with the first sample feature, and the second sample feature sets include multiple second sample features.

[0113] S602, The evaluation model is trained based on the features of the first sample and the feature set of the second sample to obtain the trained evaluation model.

[0114] In this embodiment of the disclosure, a training dataset is obtained, wherein the training dataset includes a first sample feature and N second sample feature sets associated with the first sample feature, and then the evaluation model is trained based on the first sample feature and the second sample feature sets to obtain a trained evaluation model.

[0115] In some embodiments, S602, the evaluation model is trained based on the first sample features and the second sample feature set to obtain a trained evaluation model, including:

[0116] Based on the first sample features and the second sample features in each second sample feature set, as well as the evaluation results of the first type of samples corresponding to the first sample features and the second sample features in each second sample feature set, the evaluation model is trained to obtain the trained first type of evaluation model corresponding to the first sample features and each second sample feature set.

[0117] Based on the first sample features and the second sample features in any two sets of second sample features, as well as the evaluation results of the second type of samples corresponding to the first sample features and the second sample features in any two sets of second sample features, the evaluation model is trained to obtain the trained second type of evaluation model corresponding to the first sample features and any two sets of second sample features.

[0118] Similarly, based on the first sample features and the second sample features in any M sets of second sample features, as well as the evaluation results of the Mth class of samples corresponding to the first sample features and the second sample features in any M sets of second sample features, the evaluation model is trained to obtain the trained Mth class evaluation model corresponding to the first sample features and any M sets of second sample features.

[0119] In this embodiment of the disclosure, the evaluation model is trained based on the first sample features and the second sample features in each of the second sample feature sets, as well as the evaluation results of the first type of samples corresponding to the first sample features and the second sample features in each of the second sample feature sets, to obtain the trained first type of evaluation model corresponding to the first sample features and each of the second sample feature sets; the evaluation model is trained based on the first sample features and the second sample features in any two of the second sample feature sets, as well as the evaluation results of the second type of samples corresponding to the first sample features and the second sample features in any two of the second sample feature sets, to obtain the trained second type of evaluation model corresponding to the first sample features and any two of the second sample feature sets; and so on, the evaluation model is trained based on the first sample features and the second sample features in any M of the second sample feature sets, as well as the evaluation results of the Mth type of samples corresponding to the first sample features and the second sample features in any M of the second sample feature sets, to obtain the trained Mth type of evaluation model corresponding to the first sample features and any M of the second sample feature sets.

[0120] In some embodiments, the first-class sample evaluation results corresponding to the first sample feature and the second sample features in each second sample feature set can be determined by means of manual annotation.

[0121] In some embodiments, the second-class sample evaluation results corresponding to the first sample feature and the second sample features in any two second sample features can be determined by means of manual annotation.

[0122] In some embodiments, the M-class sample evaluation results corresponding to the first sample feature and the second sample features in any M second sample features can be determined by means of manual annotation.

[0123] Exemplarily, in the case of determining a first sample feature and two second sample feature sets, as well as the first-class sample evaluation results corresponding to the first sample feature and the second sample features in the first second sample feature set, and the first-class sample evaluation results corresponding to the first sample feature and the second sample features in the second second sample feature set, the evaluation model is trained to obtain a trained first-class evaluation model corresponding to the first sample feature and each second sample feature set.

[0124] Exemplarily, for M second sample feature sets, 2sqrt(M) evaluation models [Model1, Model2,... Model2sqrt(M)] are constructed, and the evaluation model can be a multi-class radial basis kernel support vector machine SVM (Support Vector Machine) model.

[0125] For each evaluation model, the number P (P < M) of second sample feature sets is randomly selected for it, and P second sample feature sets are randomly selected: X = [x1, x2, x3... xP].

[0126] First, use a linear transformation function to map the P second sample feature sets into a linearly separable space in a high dimension, and the relationship between this linear transformation and the selected radial basis kernel function is as follows:

[0127]

[0128] In the high-dimensional separable space, the optimization problem of the support vector machine is:

[0129]

[0130] By implementing the embodiments of the present disclosure, the efficiency of screening achievement transformation data can be improved, and the accuracy of data screening can be improved when performing data screening based on the evaluation results.

[0131] Figure 7This is a structural diagram of a data screening device 10 for technology transfer provided in an embodiment of this disclosure.

[0132] like Figure 7 As shown, the data screening device 10 for the transformation of achievements includes: a feature extraction unit 11, a model determination unit 12, and a model prediction unit 13.

[0133] The feature extraction unit 11 is used to input the technical text to be evaluated into the corresponding knowledge graph for feature extraction, and to obtain a first feature and N second features associated with the first feature, where N is an integer greater than 0.

[0134] The model determination unit 12 is used to determine the corresponding trained evaluation model based on the first feature and the second feature.

[0135] The model prediction unit 13 is used to determine the evaluation result of the potential for commercialization of the technical text to be evaluated based on the first feature, the second feature, and the trained evaluation model.

[0136] In some embodiments, when N is an integer greater than 1, the model determination unit 12 is specifically used for:

[0137] Based on the first feature and each of the second features, determine the corresponding trained first-class evaluation model;

[0138] Based on the first feature and any two second features, determine the corresponding trained second-class evaluation model;

[0139] Similarly, based on the first feature and any M second features, the corresponding trained M-th class evaluation model is determined, where M is an integer greater than 1 and less than or equal to N.

[0140] In some embodiments, the model prediction unit 13 is specifically used for:

[0141] Input the first feature and each second feature into the corresponding trained first-class evaluation model to obtain N first-class prediction results, and determine the first evaluation value based on the N first-class prediction results;

[0142] Input the first feature and any two second features into the corresponding trained second-class evaluation model to obtain... The second type of prediction result, and based on Based on the second type of prediction results, determine the second evaluation value;

[0143] Similarly, by inputting the first feature and any M second features into the corresponding trained M-th class evaluation model, we obtain... The prediction result of the Mth class, and based on Given the Mth type prediction result, determine the Mth evaluation value;

[0144] Based on the first evaluation value, the second evaluation value, and the Mth evaluation value, the evaluation result is determined to indicate whether the technical text to be evaluated has the potential for commercialization.

[0145] like Figure 8 As shown, in some embodiments, the achievement transformation data screening device 10 further includes: a category determination unit 14 and a knowledge graph determination unit 15.

[0146] Category determination unit 14 is used to determine the technology category corresponding to the text to be evaluated.

[0147] The knowledge graph determination unit 15 is used to determine the knowledge graph corresponding to the technology text to be evaluated based on the technology category.

[0148] like Figure 9 As shown, in some embodiments, the achievement transformation data screening device 10 further includes: a sample data acquisition unit 16 and a knowledge graph construction unit 17.

[0149] The sample data acquisition unit 16 is used to acquire sample text data of different technical categories.

[0150] Knowledge graph construction unit 17 is used to construct a knowledge graph of the technology category based on sample text data.

[0151] In some embodiments, the knowledge graph construction unit 17 is specifically used for:

[0152] Identify the sample entities in the sample text data;

[0153] Retrieve associated data from the sample text data that is related to the sample entities;

[0154] The associated data is preprocessed to obtain a set of reference sample entities associated with the sample entity, wherein the set of reference sample entities includes at least one reference associated entity;

[0155] Based on the correlation between the reference related entities and the sample entities, and the preset filtering dimensions, the reference sample entities in the reference related entities are determined.

[0156] A knowledge graph is constructed based on sample entities and reference sample entities.

[0157] like Figure 10 As shown, in some embodiments, the data screening device 10 for the transformation of results further includes a dataset acquisition unit 18 and a model training unit 19.

[0158] The dataset acquisition unit 18 is used to acquire the training dataset, wherein the training dataset includes a first sample feature and N second sample feature sets associated with the first sample feature, and the second sample feature sets include multiple second sample features.

[0159] The model training unit 19 is used to train the evaluation model based on the features of the first sample and the feature set of the second sample to obtain the trained evaluation model.

[0160] In some embodiments, the model training unit 19 is specifically used for:

[0161] Based on the first sample features and the second sample features in each second sample feature set, as well as the evaluation results of the first type of samples corresponding to the first sample features and the second sample features in each second sample feature set, the evaluation model is trained to obtain the trained first type of evaluation model corresponding to the first sample features and each second sample feature set.

[0162] Based on the first sample features and the second sample features in any two sets of second sample features, as well as the evaluation results of the second type of samples corresponding to the first sample features and the second sample features in any two sets of second sample features, the evaluation model is trained to obtain the trained second type of evaluation model corresponding to the first sample features and any two sets of second sample features.

[0163] Similarly, based on the first sample features and the second sample features in any M sets of second sample features, as well as the evaluation results of the Mth class of samples corresponding to the first sample features and the second sample features in any M sets of second sample features, the evaluation model is trained to obtain the trained Mth class evaluation model corresponding to the first sample features and any M sets of second sample features.

[0164] It should be noted that the foregoing explanation of the data screening method for technology transfer also applies to the data screening device for technology transfer in this embodiment of the present disclosure, and will not be repeated here.

[0165] It should be noted that the beneficial effects achieved by the technology transfer data screening device in this embodiment are the same as those achieved by the aforementioned technology transfer data screening method, and will not be repeated here.

[0166] According to embodiments of this disclosure, an electronic device is also provided, including: a processor; and a memory for storing processor-executable instructions, wherein the processor is configured to: implement the results transformation data screening method disclosed in embodiments of this disclosure.

[0167] To implement the above embodiments, this disclosure also proposes a storage medium.

[0168] When the instructions in the storage medium are executed by the processor, the processor is able to execute the results transformation data filtering method disclosed in this embodiment.

[0169] To implement the above embodiments, this disclosure also provides a computer program product.

[0170] When the computer program product is executed by the processor of the electronic device, the electronic device is able to execute the results transformation data screening method disclosed in the embodiments of this disclosure.

[0171] Figure 11 This is a structural block diagram of an electronic device according to an exemplary embodiment. Figure 10 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0172] like Figure 10 As shown, the electronic device 1000 includes a processor 111, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 112 or a program loaded from memory 116 into random access memory (RAM) 113. The RAM 113 also stores various programs and data required for the operation of the electronic device 1000. The processor 111, ROM 112, and RAM 113 are interconnected via a bus 114. An input / output (I / O) interface 115 is also connected to the bus 114.

[0173] The following components are connected to I / O interface 115: memory 116 including hard disks, etc.; and communication section 117 including network interface cards such as local area network (LAN) cards, modems, etc., communication section 117 performs communication processing via a network such as the Internet; and driver 118 is also connected to I / O interface 115 as needed.

[0174] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program carried on a computer-readable medium, the computer program containing program code for performing the results transformation data screening method shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 117. When the computer program is executed by the processor 111, it performs the functions defined in the results transformation data screening method of this disclosure.

[0175] In an exemplary embodiment, a storage medium including instructions is also provided, such as a memory including instructions, which can be executed by the processor 111 of the electronic device 1000 to perform the above-described method. Optionally, the storage medium may be a non-transitory computer-readable storage medium, such as a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device.

[0176] In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wireline, optical fiber, RF, etc., or any suitable combination thereof.

[0177] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0178] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A method for screening commercialization data, characterized in that, include: The technology category corresponding to the text to be evaluated is determined, and the knowledge graph corresponding to the technology text to be evaluated is determined based on the technology category. The texts to be evaluated belonging to different technology categories correspond to different knowledge graphs. The text of the technology to be evaluated is input into the corresponding knowledge graph for feature extraction to obtain a first feature and N second features associated with the first feature, where N is an integer greater than 0; When N is an integer greater than 1, the corresponding trained first-class evaluation model is determined based on the first feature and each of the second features. Based on the first feature and any two of the second features, determine the corresponding trained second-class evaluation model; Similarly, based on the first feature and any M second features, the corresponding trained M-class evaluation model is determined, where M is an integer greater than 1 and less than or equal to N; The first feature and each of the second features are input into the corresponding trained first-class evaluation model to obtain N first-class prediction results, and the first evaluation value is determined based on the N first-class prediction results. Input the first feature and any two of the second features into the corresponding trained second-class evaluation model to obtain... The second type of prediction result, and according to the Based on the second type of prediction results, determine the second evaluation value; Similarly, the first feature and any M second features are input into the corresponding trained M-th class evaluation model to obtain... The prediction result of the Mth class, and according to the Given the Mth type prediction result, determine the Mth evaluation value; Based on the first evaluation value, the second evaluation value, and the Mth evaluation value, the evaluation result is determined to indicate the potential for commercialization of the technical text to be evaluated.

2. The method as described in claim 1, characterized in that, The method further includes: Obtain sample text data for different technology categories; A knowledge graph of the technology category to which the sample text data belongs is constructed.

3. The method as described in claim 2, characterized in that, The construction of a knowledge graph of the technology category based on the sample text data includes: Identify the sample entities in the sample text data; Obtain the associated data related to the sample entity from the sample text data; The associated data is preprocessed to obtain a set of reference sample entities associated with the sample entity, wherein the set of reference sample entities includes at least one reference associated entity; Based on the correlation between the reference associated entity and the sample entity, and the preset filtering dimensions, the reference sample entity in the reference associated entity is determined; The knowledge graph is constructed based on the sample entities and the reference sample entities.

4. The method according to any one of claims 1 to 3, characterized in that, The method further includes: Obtain a training dataset, wherein the training dataset includes a first sample feature and N sets of second sample features associated with the first sample feature, and the sets of second sample features include multiple second sample features; The evaluation model is trained based on the first sample features and the second sample feature set to obtain a trained evaluation model.

5. The method as described in claim 4, characterized in that, The step of training the evaluation model based on the first sample features and the second sample feature set to obtain the trained evaluation model includes: The evaluation model is trained based on the first sample feature and the second sample feature in each of the second sample feature sets, as well as the evaluation results of the first type of samples corresponding to the first sample feature and the second sample feature in each of the second sample feature sets, to obtain the trained first type of evaluation model corresponding to the first sample feature and each of the second sample feature sets. The evaluation model is trained based on the first sample feature and the second sample feature in any two sets of the second sample feature, as well as the evaluation results of the second type of sample corresponding to the first sample feature and the second sample feature in any two sets of the second sample feature, to obtain the trained second type of evaluation model corresponding to the first sample feature and any two sets of the second sample feature. Similarly, based on the first sample feature and the second sample features in any M sets of second sample features, as well as the evaluation results of the Mth class of samples corresponding to the first sample feature and the second sample features in any M sets of second sample features, the evaluation model is trained to obtain the trained Mth class evaluation model corresponding to the first sample feature and any M sets of second sample features.

6. A data screening device for technology transfer, characterized in that, include: The feature extraction unit is used to determine the technology category corresponding to the text to be evaluated, and to determine the knowledge graph corresponding to the technology text to be evaluated based on the technology category. The text to be evaluated belongs to different technology categories and corresponds to different knowledge graphs. The text to be evaluated is input into the corresponding knowledge graph for feature extraction to obtain a first feature and N second features associated with the first feature, where N is an integer greater than 0. The model determination unit is used to determine the corresponding trained evaluation model based on the first feature and the second feature; The model prediction unit is used to determine the evaluation result of the possibility of commercialization of the technical text to be evaluated based on the first feature, the second feature and the trained evaluation model. The model determination unit is specifically used for: When N is an integer greater than 1, the corresponding trained first-class evaluation model is determined based on the first feature and each of the second features. Based on the first feature and any two of the second features, determine the corresponding trained second-class evaluation model; Similarly, based on the first feature and any M second features, the corresponding trained M-class evaluation model is determined, where M is an integer greater than 1 and less than or equal to N; The model prediction unit is specifically used for: The first feature and each of the second features are input into the corresponding trained first-class evaluation model to obtain N first-class prediction results, and the first evaluation value is determined based on the N first-class prediction results. Input the first feature and any two of the second features into the corresponding trained second-class evaluation model to obtain... The second type of prediction result, and according to the Based on the second type of prediction results, determine the second evaluation value; Similarly, the first feature and any M second features are input into the corresponding trained M-th class evaluation model to obtain... The prediction result of the Mth class, and according to the Given the Mth type prediction result, determine the Mth evaluation value; Based on the first evaluation value, the second evaluation value, and the Mth evaluation value, the evaluation result is determined to indicate that the technical text to be evaluated has the potential for commercialization.

7. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 5.

Images