A method for accurate recommendation of scientific and technological achievements

Abstract

The present invention discloses a method for accurately recommending scientific and technological achievements, which belongs to the technical field of scientific and technological achievement recommendation and solves the problem of low accuracy of existing scientific and technological achievement recommendation results. The method includes: performing technical expansion and correlation analysis on the demand for scientific and technological achievements of an enterprise to obtain the technical expansion demand for the demand for scientific and technological achievements of the enterprise; constructing a knowledge graph based on the demand for scientific and technological achievements of the enterprise and its technical expansion demand, and obtaining the representation vector of the demand for scientific and technological achievements of the enterprise through the graph representation method; constructing a key feature matrix of the demand for scientific and technological achievements of the enterprise; calculating the coverage degree of the key feature matrix of each scientific and technological achievement in the scientific and technological achievement library to the key feature matrix of the demand for scientific and technological achievements of the enterprise, and taking the scientific and technological achievements whose coverage degree exceeds the coverage degree threshold as the candidate scientific and technological achievements; inputting the representation vector of the demand for scientific and technological achievements of the enterprise and the candidate scientific and technological achievements into the prediction model, and the prediction model selects the target scientific and technological achievements from the candidate scientific and technological achievements and recommends them to the corresponding enterprise.

Description

Technical Field

[0001] The present invention relates to the technical field of scientific and technological achievement recommendation, and in particular to a method for accurately recommending scientific and technological achievements. Background Art

[0002] As an important means and tool for science and technology management, the recommendation of scientific and technological achievements plays an important role in the transformation and transaction of scientific and technological achievements, effectively promoting my country's economic and social development and scientific and technological progress, and is an important part of scientific and technological activities and management. With the accelerated evolution of a new round of scientific and technological revolution and industrial transformation, the complexity and uncertainty of scientific and technological activities have increased, and higher requirements have been placed on the accuracy and effectiveness of the recommendation of scientific and technological achievements.

[0003] At present, the method for enterprises to collect scientific and technological achievements mainly relies on the staff in the relevant technical fields to manually collect them through the website, and then screen the collected scientific and technological achievements. This method not only takes a lot of time, but also requires a lot of human resources. In addition, the staff cannot accurately screen out the scientific and technological achievements needed by the enterprise during the screening process. In addition, the scientific and technological achievement recommendation platform also proposed a method for intelligent recommendation of scientific and technological achievements, which mainly compares the keywords of scientific and technological achievements and enterprise needs, and uses the comparison results as the criteria for intelligent recommendation of scientific and technological achievements.

[0004] However, the development speed of scientific and technological achievements is very fast now. Some emerging technologies are not within the current research scope of enterprises, but have guiding significance for the future development of enterprises. Such scientific and technological achievements are difficult to recommend to customers through keyword comparison, making it difficult to achieve accurate recommendation of scientific and technological achievements. Summary of the invention

[0005] In view of the above analysis, an embodiment of the present invention aims to provide a method for accurately recommending scientific and technological achievements, so as to solve the problem of low accuracy of scientific and technological achievement recommendation results obtained by existing manual collection methods and keyword comparison methods.

[0006] The present invention discloses a method for accurately recommending scientific and technological achievements, the method comprising:

[0007] Obtain the demand for scientific and technological achievements of enterprises, conduct technical expansion and correlation analysis on the demand for scientific and technological achievements of enterprises, and obtain the technical expansion demand for scientific and technological achievements of enterprises;

[0008] Based on the enterprise's demand for scientific and technological achievements and its technology expansion needs, a knowledge graph of the enterprise's demand for scientific and technological achievements is constructed, and then the representation vector of the enterprise's demand for scientific and technological achievements is obtained through the graph representation method;

[0009] Extract the scientific and technological achievement demand phrases from the enterprise's scientific and technological achievement demand and its technology expansion demand; and based on the scientific and technological achievement demand phrases, construct a key feature matrix of the enterprise's scientific and technological achievement demand;

[0010] Calculate the coverage of the key feature matrix of each scientific and technological achievement in the scientific and technological achievement database to the key feature matrix of the enterprise's scientific and technological achievement needs, and take the scientific and technological achievements whose coverage exceeds the coverage threshold as candidate scientific and technological achievements;

[0011] The representation vectors of the enterprise's scientific and technological achievement needs and alternative scientific and technological achievements are input into the trained prediction model, and the trained prediction model selects the target scientific and technological achievements from the alternative scientific and technological achievements and recommends them to the corresponding enterprises.

[0012] On the basis of the above scheme, the present invention also makes the following improvements:

[0013] Furthermore, the technical expansion and correlation analysis of the enterprise's scientific and technological achievement needs are carried out:

[0014] Obtain the scientific and technological entity words, common replacement words for scientific and technological entity words, and demand description sentences containing scientific and technological entity words for the enterprise's scientific and technological achievements;

[0015] Use the generative artificial intelligence interaction system to expand the demand for enterprise scientific and technological achievements and obtain multiple artificial intelligence interaction statements;

[0016] Select AI interaction sentences that contain scientific and technological entity words or common replacement words for scientific and technological entity words and are technical sentences as sentences to be analyzed, and select sentences to be analyzed that have a strong correlation with demand description sentences as technical extension sentences for enterprise scientific and technological achievement requirements;

[0017] Summarize all technical extension statements that have a strong correlation with the requirement description statements as the technical extension requirements for the enterprise's scientific and technological achievements.

[0018] Further, the statement to be analyzed that has a strong correlation with the demand description statement is used as a technical extension statement of the enterprise's scientific and technological achievement demand, and is executed:

[0019] Use the dependency syntax parser to parse each requirement description statement and statement to be analyzed, generate the corresponding dependency tree, and extract the corresponding dependency relationship from the dependency tree;

[0020] For the requirement description statement and the statement to be analyzed that contain the same scientific and technological entity word, the requirement description statement is used as a benchmark, and the statement to be analyzed whose dependency relationship is similar or consistent with the requirement description statement is selected for semantic similarity judgment. If the semantic similarity meets the preset similarity threshold, the statement to be analyzed is regarded as a technical extension statement that has a strong correlation with the requirement description statement.

[0021] Furthermore, the key feature matrix of the enterprise's scientific and technological achievement needs is constructed by:

[0022] The demand description statements and technical extension statements of the enterprise's scientific and technological achievements are reconstructed according to the preset phrase rules to obtain the scientific and technological achievements demand phrases of the corresponding statements;

[0023] Convert the scientific and technological achievement requirement phrases into corresponding feature vectors.

[0024] The feature vectors of each scientific and technological achievement demand phrase are used as elements of the key feature matrix of the enterprise's scientific and technological achievement demand, and the key feature matrix of the enterprise's scientific and technological achievement demand is constructed.

[0025] Further, the scientific and technological achievement requirement phrase of the corresponding statement is obtained by executing:

[0026] The demand description sentences and technical extension sentences of the enterprise's scientific and technological achievements are segmented and stop words are removed, and the entity information in the corresponding sentences is extracted through the entity annotation model;

[0027] According to the preset phrase rules, the entity information in the corresponding sentence is reconstructed to obtain the scientific and technological achievement requirement phrase of the corresponding sentence.

[0028] Furthermore, the trained prediction model selects target scientific and technological achievements from candidate scientific and technological achievements and recommends them to corresponding enterprises, and executes:

[0029] Use the trained prediction model to predict the demand for enterprise scientific and technological achievements, and calculate the matching score between each candidate scientific and technological achievement and the demand for enterprise scientific and technological achievements;

[0030] Sort candidate scientific and technological achievements according to the matching scores output by the prediction model;

[0031] According to the ranking results, several scientific and technological achievements with the highest scores are selected as target scientific and technological achievements and recommended to corresponding enterprises.

[0032] Furthermore, the prediction model is trained in the following way:

[0033] Acquire multiple sets of training sample data, each set of sample data includes enterprise scientific and technological achievement demand samples, scientific and technological achievement samples, and matching scores between each scientific and technological achievement sample and the corresponding enterprise scientific and technological achievement demand sample;

[0034] Obtaining the characterization vectors of enterprise scientific and technological achievement demand samples and scientific and technological achievement samples in the sample data; inputting the characterization vectors of enterprise scientific and technological achievement demand samples into the prediction model, and using them to train and fit the matching scores of the characterization vectors of the corresponding scientific and technological achievement samples;

[0035] The training is completed directly and a trained prediction model is obtained.

[0036] Furthermore, the scientific and technological achievement samples include both positive scientific and technological achievement samples and negative scientific and technological achievement samples.

[0037] Furthermore, for each scientific and technological achievement, a representation vector of the scientific and technological achievement is constructed and executed:

[0038] The corresponding knowledge graph is constructed based on the scientific and technological achievements, and then the representation vector of the scientific and technological achievements is obtained through the graph representation method.

[0039] Furthermore, for each scientific and technological achievement, a key feature matrix of the scientific and technological achievement is constructed and executed:

[0040] Extract scientific and technological entity words from scientific and technological achievements, and regard sentences containing any scientific and technological entity words in scientific and technological achievements as scientific and technological achievement sentences containing scientific and technological entity words;

[0041] Reconstruct the scientific and technological achievement sentences of the scientific and technological achievements according to the preset phrase rules to obtain the scientific and technological achievement phrases of the corresponding sentences;

[0042] The scientific and technological achievement phrases are converted into corresponding feature vectors, and the feature vectors of each scientific and technological achievement phrase are used as elements of the key feature matrix of the scientific and technological achievements to construct the key feature matrix of the corresponding scientific and technological achievements.

[0043] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0044] The present invention provides a method for accurately recommending scientific and technological achievements. First, the technical expansion requirements of the enterprise's scientific and technological achievement requirements are obtained by performing technical expansion and correlation analysis on the enterprise's scientific and technological achievement requirements, and a representation vector is constructed accordingly, thereby enriching the content of the requirements and improving the accuracy of the recommendation results. The enterprise's scientific and technological achievement requirements and their technical expansion requirements are used to obtain scientific and technological achievement entity words, commonly used replacement words, and demand description sentences, and a key feature matrix of the enterprise's scientific and technological achievement requirements is constructed; and the key feature matrix of each scientific and technological achievement in the scientific and technological achievement library is used to determine the coverage of the key feature matrix of the enterprise's scientific and technological achievement requirements, thereby effectively reducing the recommended scope of scientific and technological achievements, reducing the prediction difficulty and complexity of the later prediction model, and facilitating a more accurate prediction effect. On this basis, the representation vectors of the enterprise's scientific and technological achievement requirements and the alternative scientific and technological achievements are input into the trained prediction model, and the trained prediction model selects the target scientific and technological achievements from the alternative scientific and technological achievements and recommends them to the corresponding enterprise, thereby realizing accurate recommendation of scientific and technological achievements.

[0045] The method for accurately recommending scientific and technological achievements provided by the present invention has clear logic, strong adaptability, small amount of calculation, high recommendation accuracy, and good promotion value and application prospect.

[0046] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings are only used for the purpose of illustrating specific embodiments and are not to be considered as limiting the present invention. In the entire drawings, the same reference symbols represent the same components;

[0048] Figure 1 A flowchart of a method for accurately recommending scientific and technological achievements provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0050] A specific embodiment of the present invention discloses a method for accurately recommending scientific and technological achievements, the flow chart of which is as follows: Figure 1 As shown, the method includes the following steps.

[0051] Step S1: Obtain the enterprise's demand for scientific and technological achievements, conduct technical expansion and correlation analysis on the enterprise's demand for scientific and technological achievements, and obtain the technical expansion demand of the enterprise's demand for scientific and technological achievements.

[0052] In this embodiment, the enterprise's demand for scientific and technological achievements includes but is not limited to the technical direction and content of the enterprise's needs, the enterprise's development direction, technical shortcomings, R&D investment and industrial chain situation. In the specific implementation process, the content of the enterprise's demand for scientific and technological achievements can also be enriched by adding content related to the enterprise's demand for scientific and technological achievements, such as future R&D directions and future R&D ideas. Preferably, in order to avoid the limitations of the writing of the enterprise's demand for scientific and technological achievements, in this embodiment, the following operations are performed to achieve technical expansion and correlation analysis of the enterprise's demand for scientific and technological achievements:

[0053] Step S11: Obtain the scientific and technological entity words of the enterprise's scientific and technological achievement requirements, the commonly used replacement words of the scientific and technological entity words, and the requirement description sentences containing the scientific and technological entity words.

[0054] In the specific implementation process, the enterprise directly specifies the scientific and technological entity words when writing the enterprise's scientific and technological achievement requirements, or determines the scientific and technological entity words in the enterprise's scientific and technological achievement requirements by matching the scientific and technological achievement entity library. After that, the sentence containing any scientific and technological entity words in the enterprise's scientific and technological achievement requirements is used as the requirement description sentence.

[0055] In this embodiment, common replacement words for scientific and technological entity words can be obtained by searching for their abbreviations, English translations, abbreviations, aliases, etc., thereby avoiding the problem of poor recommendation accuracy caused by missing common replacement words for scientific and technological entity words.

[0056] Step S12: Utilize the generative artificial intelligence interaction system to expand the enterprise's demand for scientific and technological achievements and obtain multiple artificial intelligence interaction statements.

[0057] Preferably, in the process of expanding the enterprise's demand for scientific and technological achievements using a generative artificial intelligence interaction system, in order to ensure the expansion effect, the input template of the generative artificial intelligence interaction system can be preset according to the enterprise's demand for scientific and technological achievements, such as: introducing the scientific and technological achievements of the "enterprise's demand for scientific and technological achievements" in a certain field. In addition, keyword restrictions can be added to the preset input template of the generative artificial intelligence interaction system. The keywords include but are not limited to the above scientific and technological entity words and their commonly used replacement words. Such as: search for scientific and technological achievements related to the "enterprise's demand for scientific and technological achievements" and containing the following "keywords". It should be noted that since this embodiment aims to achieve scientific and technological achievement recommendations, the training corpus of the generative artificial intelligence interaction system should contain a large amount of scientific and technological achievement knowledge. Common generative artificial intelligence interaction systems include ChatGPT, Wenxinyiyan or chatGLM.

[0058] Step S13: Select artificial intelligence interactive sentences that contain scientific and technological entity words or commonly used replacement words for scientific and technological entity words and are technical sentences as sentences to be analyzed, and use sentences to be analyzed that have a strong correlation with demand description sentences as technical extension sentences of the enterprise's scientific and technological achievement requirements.

[0059] After obtaining multiple artificial intelligence interaction statements, it is judged whether each artificial intelligence interaction statement is a technical statement, and non-technical statements are eliminated to obtain the remaining technical statements. In the specific implementation process, the judgment of whether a technical statement is a technical statement or not can be realized by a technical classifier to realize the binary classification judgment of "yes / no is a technical statement" for the artificial intelligence interaction statement. Preferably, the technical classifier can be obtained based on neural network training, and its main purpose is to perform a binary classification judgment of "yes / no is a technical statement" on the input information. If so, the judgment result of the technical classifier is "the input statement is a technical statement"; if not, the judgment result of the technical classifier is "the input statement is not a technical statement". Exemplarily, the technical classifier can be obtained based on a general CNN neural network model. In the specific implementation process, a large number of input statements and their technical labels can be prepared in advance (exemplarily, 0 and 1 labels can be designed, where 0 represents that the input statement is a technical statement and 1 represents that the input statement is not a technical problem) to form a technical statement training sample set. And based on the technical statement training sample set, the neural network corresponding to the technical classifier is trained for multiple rounds. After the training is passed, a technical classifier that can be used for actual technical statement judgment is obtained.

[0060] On the basis of obtaining the remaining technical sentences, artificial intelligence interactive sentences containing scientific and technological entity words or common replacement words for scientific and technological entity words are selected as sentences to be analyzed. Preferably, in this embodiment, the sentences to be analyzed that have a strong correlation with the demand description sentence are used as technical extension sentences of the enterprise's scientific and technological achievement requirements, and the following are executed:

[0061] Step S131: Use the dependency syntax parser to parse each requirement description statement and statement to be analyzed respectively, generate a corresponding dependency tree, and extract the corresponding dependency relationship from the dependency tree, including the dominance relationship between words (such as subject-predicate relationship, verb-object relationship, etc.).

[0062] Step S132: For the requirement description statement and the statement to be analyzed that contain the same scientific and technological entity word, the requirement description statement is used as a benchmark, and the statement to be analyzed whose dependency relationship is similar or consistent with the requirement description statement is selected for semantic similarity judgment. If the semantic similarity meets the preset similarity threshold, the statement to be analyzed is regarded as a technical extension statement that has a strong correlation with the requirement description statement.

[0063] It should be noted that dependency refers to the logical and semantic connection between words in a sentence. The process of selecting sentences to be analyzed whose dependency relationship is similar or consistent with the requirement description sentence is implemented by combining one or more of the following comparison methods: Word meaning comparison: compare whether the keywords in the two sentences are the same or have similar word meanings. Syntactic structure comparison: analyze whether the syntactic structures of the two sentences are consistent, such as subject-predicate structure, verb-object structure, etc. Dependency syntactic analysis: use dependency syntactic analysis tools to determine the relationship between the various components in the sentence, such as subject, predicate, object, attributive, adverbial, etc. The specific judgment criteria for sentence similarity or consistency can be set according to the specific situation. For example, when a comparison method is used to judge the semantic similarity, the comparison result of the corresponding comparison method can be compared with the similarity threshold. If the comparison result is 1, it is judged that the semantics are consistent; if the comparison result is between the similarity threshold and 1, it is judged that the sentences are similar. When multiple comparison methods are used for combined judgment, the comparison result of each comparison method is required to be greater than or equal to the similarity threshold.

[0064] Step S14: Summarize all technical extension statements that have a strong correlation with the requirement description statement as the technical extension requirements of the enterprise's scientific and technological achievement requirements.

[0065] Step S2: Based on the enterprise's demand for scientific and technological achievements and its technical expansion needs, a knowledge graph of the enterprise's demand for scientific and technological achievements is constructed, and then a representation vector of the enterprise's demand for scientific and technological achievements is obtained through a graph representation method.

[0066] Step S21: extract entities (such as technology, products and services), entity attributes (such as technical parameters, market size) and relationships between entities (such as demand, application field) from the enterprise's demand for scientific and technological achievements and its technology expansion needs.

[0067] Step S22: Based on the extracted entities, relationships and attributes, a knowledge graph of the enterprise's scientific and technological achievement needs is constructed.

[0068] During the specific implementation process, a knowledge graph can be built using a graph database, such as Neo4j, to store and manage graph data.

[0069] Step S23: Use a graph embedding algorithm (such as Node2Vec, GraphSAGE or TransE) to represent the knowledge graph and obtain a representation vector of the enterprise's demand for scientific and technological achievements.

[0070] Step S3: Extract the scientific and technological achievement demand phrases in the enterprise's scientific and technological achievement demand and its technology expansion demand; and construct a key feature matrix of the enterprise's scientific and technological achievement demand based on the scientific and technological achievement demand phrases.

[0071] Based on the requirement description statement and technology extension statement of the enterprise's scientific and technological achievement requirement obtained in step S1, execute:

[0072] Step S31: reconstructing the requirement description statement and the technical extension statement of the enterprise's scientific and technological achievement requirement according to preset phrase rules to obtain the scientific and technological achievement requirement phrases of the corresponding statements.

[0073] For example, the requirement description statement and the technical extension statement of the enterprise's scientific and technological achievement requirement can be reconstructed according to the preset phrase rules in the following manner:

[0074] Step S311: perform word segmentation and stop word removal on the demand description statements and technology extension statements of the enterprise's scientific and technological achievement requirements, and extract entity information in the corresponding statements through the entity annotation model.

[0075] In the specific implementation process, the entity annotation model based on GlobalPointer in the prior art can be used to implement entity annotation. This embodiment mainly involves the application of the model, and the specific implementation process of the model will not be repeated.

[0076] Exemplarily, the entity information extracted from the corresponding sentence may include the following categories: scientific and technological entity words, efficacy words (used to describe functional effects), verbs, parameters, and scientific and technological entity word adjectives.

[0077] Step S312: reconstruct the entity information in the corresponding sentence according to the preset phrase rules to obtain the scientific and technological achievement requirement phrase of the corresponding sentence.

[0078] In this embodiment, the purpose of the preset phrase rules is to extract the core meaning of the corresponding sentence, so as to quickly determine the core content of the sentence. In the specific implementation process, one or more phrase rules are selected for reconstruction according to the specific application scenario. For example, the preset phrase rules can be: (1) scientific and technological entity word + functional word + parameter; (2) scientific and technological entity word + verb + parameter; (3) scientific and technological entity word + scientific and technological entity word adjective, etc.

[0079] Step S32: converting the scientific and technological achievement requirement phrase into a corresponding feature vector.

[0080] According to the word vector of the entity information in the phrase of scientific and technological achievement demand, the phrase vector of the phrase of scientific and technological achievement demand is calculated, and the phrase vector of the phrase of scientific and technological achievement demand is used as the feature vector of the corresponding sentence.

[0081] In the specific implementation process, the word2vec model can be used to calculate the word vector of the entity information in the scientific and technological achievement requirement phrase. The word vector of the entity information in the scientific and technological achievement requirement phrase is weighted summed to calculate the phrase vector of the scientific and technological achievement requirement phrase.

[0082] Step S33: Taking the feature vectors of each scientific and technological achievement demand phrase as the elements of the key feature matrix of the enterprise's scientific and technological achievement demand, a key feature matrix of the enterprise's scientific and technological achievement demand is constructed.

[0083] Step S4: Calculate the coverage of the key feature matrix of each scientific and technological achievement in the scientific and technological achievement database to the key feature matrix of the enterprise's scientific and technological achievement needs, and take the scientific and technological achievements whose coverage exceeds the coverage threshold as candidate scientific and technological achievements.

[0084] For example, if the key feature matrix of a scientific and technological achievement covers all the feature vectors in the key feature matrix of the enterprise's scientific and technological achievement requirements, the coverage is 100%. If the key feature matrix of a scientific and technological achievement does not contain any feature vector in the key feature matrix of the enterprise's scientific and technological achievement requirements, the coverage is 0%.

[0085] By determining the coverage, scientific and technological achievements that are highly correlated with the needs of the enterprise's scientific and technological achievements can be pre-screened as candidate scientific and technological achievements, which is convenient for later screening out target scientific and technological achievements that are more suitable for the needs of the enterprise's scientific and technological achievements from the candidate scientific and technological achievements, thereby achieving accurate recommendation of scientific and technological achievements. At the same time, this method effectively narrows the scope of scientific and technological achievements predicted by the model, reduces the difficulty and complexity of the prediction model in the later stage, and facilitates obtaining more accurate prediction results.

[0086] Step S5: Input the characterization vectors of the enterprise's scientific and technological achievement needs and the candidate scientific and technological achievements into the trained prediction model, and use the trained prediction model to select the target scientific and technological achievements from the candidate scientific and technological achievements and recommend them to the corresponding enterprises.

[0087] In the specific implementation process, the trained prediction model is used to predict the demand for enterprise scientific and technological achievements, and the matching score between each candidate scientific and technological achievement and the enterprise's scientific and technological achievement demand is calculated. The candidate scientific and technological achievements are sorted according to the matching score output by the prediction model. According to the sorting results, several scientific and technological achievements with the highest scores are selected as target scientific and technological achievements and recommended to the corresponding enterprises.

[0088] Below, we introduce the training process of the prediction model:

[0089] Multiple sets of training sample data are obtained, each set of sample data includes enterprise scientific and technological achievement demand samples, scientific and technological achievement samples, and the matching scores of each scientific and technological achievement sample and the corresponding enterprise scientific and technological achievement demand sample. It should be noted that the scientific and technological achievement samples here include both positive scientific and technological achievement samples (recommended, with a high matching score) and negative scientific and technological achievement samples (not recommended, with a low matching score).

[0090] Obtain the characterization vectors of enterprise scientific and technological achievement demand samples and scientific and technological achievement samples in the sample data; input the characterization vectors of enterprise scientific and technological achievement demand samples into the prediction model, and use them to train and fit the matching scores of the characterization vectors of the corresponding scientific and technological achievement samples. Direct training is completed to obtain a trained prediction model.

[0091] It should be noted that the scientific and technological achievements database in this embodiment is pre-built. When storing scientific and technological achievements in the scientific and technological achievements database, the representation vector and key feature matrix of the scientific and technological achievements can be stored in the scientific and technological achievements database at the same time. Specifically, the scientific and technological achievements database is a database of detailed information on scientific and technological achievements, including but not limited to an overview of scientific and technological achievements, technical fields, maturity, intellectual property status, awards, etc.

[0092] In this embodiment, for each scientific and technological achievement, the following operations are specifically performed to construct a representation vector of the scientific and technological achievement: a corresponding knowledge graph is constructed based on the scientific and technological achievement, and then the representation vector of the scientific and technological achievement is obtained through graph representation. The relevant implementation method is similar to step S2 and will not be repeated here.

[0093] In addition, for each scientific and technological achievement, the following operations are performed to construct a key feature matrix of the scientific and technological achievement: (1) Extract the scientific and technological entity words in the scientific and technological achievements, and take the sentences containing any scientific and technological entity words in the scientific and technological achievements as scientific and technological achievement sentences containing scientific and technological entity words. (2) Reconstruct the scientific and technological achievement sentences of the scientific and technological achievements according to the preset phrase rules to obtain the scientific and technological achievement phrases of the corresponding sentences. (3) Convert the scientific and technological achievement phrases into corresponding feature vectors, take the feature vectors of each scientific and technological achievement phrase as the elements of the key feature matrix of the scientific and technological achievements, and construct the key feature matrix of the corresponding scientific and technological achievements.

[0094] The relevant specific implementation method is similar to step S3 and will not be repeated here.

[0095] In summary, the embodiment of the present invention provides a method for accurately recommending scientific and technological achievements. First, the technical expansion requirements of the enterprise's scientific and technological achievement requirements are obtained by performing technical expansion and correlation analysis on the enterprise's scientific and technological achievement requirements, and a representation vector is constructed accordingly, thereby enriching the content of the requirements and improving the accuracy of the recommendation results. The enterprise's scientific and technological achievement requirements and their technical expansion requirements are used to obtain scientific and technological achievement entity words, commonly used replacement words, and demand description sentences, and a key feature matrix of the enterprise's scientific and technological achievement requirements is constructed; and the key feature matrix of each scientific and technological achievement in the scientific and technological achievement library is used to determine the coverage of the key feature matrix of the enterprise's scientific and technological achievement requirements to determine the candidate scientific and technological achievements, thereby effectively reducing the recommended scope of scientific and technological achievements, reducing the prediction difficulty and complexity of the later prediction model, and facilitating obtaining a more accurate prediction effect. On this basis, the representation vectors of the enterprise's scientific and technological achievement requirements and the candidate scientific and technological achievements are input into the trained prediction model, and the trained prediction model selects the target scientific and technological achievements from the candidate scientific and technological achievements and recommends them to the corresponding enterprise, thereby realizing the accurate recommendation of scientific and technological achievements. The method for accurately recommending scientific and technological achievements provided by the present invention has clear logic, strong adaptability, small amount of calculation, high recommendation accuracy, and good promotion value and application prospects.

[0096] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0097] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for accurately recommending scientific and technological achievements, characterized in that: The method comprises: Obtain the demand for scientific and technological achievements of enterprises, conduct technical expansion and correlation analysis on the demand for scientific and technological achievements of enterprises, and obtain the technical expansion demand for scientific and technological achievements of enterprises; Based on the enterprise's demand for scientific and technological achievements and its technology expansion needs, a knowledge graph of the enterprise's demand for scientific and technological achievements is constructed, and then the representation vector of the enterprise's demand for scientific and technological achievements is obtained through the graph representation method; Extract the scientific and technological achievement demand phrases from the enterprise's scientific and technological achievement demand and its technology expansion demand; and based on the scientific and technological achievement demand phrases, construct a key feature matrix of the enterprise's scientific and technological achievement demand; Calculate the coverage of the key feature matrix of each scientific and technological achievement in the scientific and technological achievement database to the key feature matrix of the enterprise's scientific and technological achievement needs, and take the scientific and technological achievements whose coverage exceeds the coverage threshold as candidate scientific and technological achievements; Input the representation vectors of the enterprise's scientific and technological achievement needs and candidate scientific and technological achievements into the trained prediction model, and use the trained prediction model to select the target scientific and technological achievements from the candidate scientific and technological achievements and recommend them to the corresponding enterprises; The above mentioned technical expansion and correlation analysis of the enterprise's scientific and technological achievements needs shall be carried out: Obtain the scientific and technological entity words, common replacement words for the scientific and technological entity words, and demand description sentences containing the scientific and technological entity words in the enterprise's demand for scientific and technological achievements; use the sentences containing any scientific and technological entity words in the enterprise's demand for scientific and technological achievements as demand description sentences; Use the generative artificial intelligence interaction system to expand the demand for enterprise scientific and technological achievements and obtain multiple artificial intelligence interaction statements; Select artificial intelligence interaction sentences that contain scientific and technological entity words or common replacement words for scientific and technological entity words and are technical sentences as sentences to be analyzed, and use sentences to be analyzed that have a strong correlation with demand description sentences as technical extension sentences of enterprise scientific and technological achievement requirements; after obtaining multiple artificial intelligence interaction sentences, judge whether each artificial intelligence interaction sentence is a technical sentence, remove non-technical sentences, and obtain remaining technical sentences; the judgment of whether it is a technical sentence or not is realized by a technical classifier to realize the binary classification judgment of "yes / no is a technical sentence" for artificial intelligence interaction sentences; on the basis of obtaining the remaining technical sentences, select artificial intelligence interaction sentences that contain scientific and technological entity words or common replacement words for scientific and technological entity words as sentences to be analyzed; Summarize all technical extension statements that have a strong correlation with the requirement description statements as the technical extension requirements of the enterprise's scientific and technological achievements; The statement to be analyzed that has a strong correlation with the demand description statement is used as a technical extension statement of the enterprise's scientific and technological achievement demand, and executes: Use the dependency syntax parser to parse each requirement description statement and statement to be analyzed, generate the corresponding dependency tree, and extract the corresponding dependency relationship from the dependency tree; For the requirement description statement and the statement to be analyzed that contain the same scientific and technological entity word, the requirement description statement is used as a benchmark, and the statement to be analyzed whose dependency relationship is similar or consistent with the requirement description statement is selected for semantic similarity judgment. If the semantic similarity meets the preset similarity threshold, the statement to be analyzed is regarded as a technical extension statement that has a strong correlation with the requirement description statement.

2. The method for accurately recommending scientific and technological achievements according to claim 1, characterized in that: The key feature matrix of enterprise scientific and technological achievement needs is constructed by: The demand description statements and technical extension statements of the enterprise's scientific and technological achievements are reconstructed according to the preset phrase rules to obtain the scientific and technological achievements demand phrases of the corresponding statements; Convert the phrases of scientific and technological achievement requirements into corresponding feature vectors; The feature vectors of each scientific and technological achievement demand phrase are used as elements of the key feature matrix of the enterprise's scientific and technological achievement demand, and the key feature matrix of the enterprise's scientific and technological achievement demand is constructed.

3. The method for accurately recommending scientific and technological achievements according to claim 2, characterized in that: To obtain the scientific and technological achievement requirement phrase of the corresponding statement, execute: The demand description sentences and technical extension sentences of the enterprise's scientific and technological achievements are segmented and stop words are removed, and the entity information in the corresponding sentences is extracted through the entity annotation model; According to the preset phrase rules, the entity information in the corresponding sentence is reconstructed to obtain the scientific and technological achievement requirement phrase of the corresponding sentence.

4. The method for accurately recommending scientific and technological achievements according to claim 3, characterized in that: The trained prediction model selects target scientific and technological achievements from candidate scientific and technological achievements and recommends them to corresponding enterprises, and executes: Use the trained prediction model to predict the demand for enterprise scientific and technological achievements, and calculate the matching score between each candidate scientific and technological achievement and the demand for enterprise scientific and technological achievements; Sort candidate scientific and technological achievements according to the matching scores output by the prediction model; According to the ranking results, several scientific and technological achievements with the highest scores are selected as target scientific and technological achievements and recommended to corresponding enterprises.

5. The method for accurately recommending scientific and technological achievements according to claim 4, characterized in that: The prediction model is trained in the following way: Acquire multiple sets of training sample data, each set of sample data includes enterprise scientific and technological achievement demand samples, scientific and technological achievement samples, and matching scores between each scientific and technological achievement sample and the corresponding enterprise scientific and technological achievement demand sample; Obtaining the characterization vectors of enterprise scientific and technological achievement demand samples and scientific and technological achievement samples in the sample data; inputting the characterization vectors of enterprise scientific and technological achievement demand samples into the prediction model, and using them to train and fit the matching scores of the characterization vectors of the corresponding scientific and technological achievement samples; The training is completed directly and a trained prediction model is obtained.

6. The method for accurately recommending scientific and technological achievements according to claim 5, characterized in that: The scientific and technological achievement samples include both positive scientific and technological achievement samples and negative scientific and technological achievement samples.

7. The method for accurately recommending scientific and technological achievements according to claim 6, characterized in that: For each scientific and technological achievement, construct a representation vector of the scientific and technological achievement and execute: The corresponding knowledge graph is constructed based on the scientific and technological achievements, and then the representation vector of the scientific and technological achievements is obtained through the graph representation method.

8. The method for accurately recommending scientific and technological achievements according to claim 7, characterized in that: For each scientific and technological achievement, a key feature matrix of the scientific and technological achievement is constructed and executed: Extract scientific and technological entity words from scientific and technological achievements, and regard sentences containing any scientific and technological entity words in scientific and technological achievements as scientific and technological achievement sentences containing scientific and technological entity words; Reconstruct the scientific and technological achievement sentences of the scientific and technological achievements according to the preset phrase rules to obtain the scientific and technological achievement phrases of the corresponding sentences; The scientific and technological achievement phrases are converted into corresponding feature vectors, and the feature vectors of each scientific and technological achievement phrase are used as elements of the key feature matrix of the scientific and technological achievements to construct the key feature matrix of the corresponding scientific and technological achievements.

Images