Intelligent recommendation method for scientific and technological achievements based on feature clustering and similarity calculation matching

Abstract

The present invention discloses a method for intelligently recommending scientific and technological achievements based on feature clustering and similarity calculation matching, including: obtaining scientific and technological achievement data, extracting key data information and text semantic information corresponding to each scientific and technological achievement; matching and processing the text semantic information corresponding to the scientific and technological achievements to obtain the fields corresponding to the scientific and technological achievements, clustering and processing the key data information corresponding to the scientific and technological achievements to obtain clustering feature words corresponding to the scientific and technological achievements, and determining the field clustering feature words corresponding to each field; constructing user demand analysis information; constructing user-related field analysis information; and determining the scientific and technological achievements recommended to the user based on the user's browsing data, the user's scientific and technological achievements, the user demand analysis information, and the user-related field analysis information. The present invention can efficiently and accurately extract corresponding scientific and technological achievements from a large amount of scientific and technological achievement data and recommend them to corresponding users, and efficiently and accurately recommend specific scientific and technological achievements to potential users.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to an intelligent recommendation method for scientific and technological achievements based on feature clustering and similarity calculation matching. Background Art

[0002] Scientific and technological achievements refer to the results of scientific research and technological development and other technological research and development activities that have certain innovation and practical value, including theoretical achievements, technical achievements and soft science achievements. Scientific and technological achievements are the core output of scientific and technological activities and play an important role in promoting scientific and technological progress and economic and social development. Scientific and technological achievements come in various forms and cover different fields.

[0003] In order to improve the efficiency and quality of scientific research, accelerate the transformation of scientific and technological achievements, and promote the deep integration of science and technology with industry, it is necessary to receive and summarize the scientific and technological achievements produced by different users, and recommend relevant scientific and technological achievements to users based on their information and needs. However, with the continuous development of technology, the number of scientific and technological achievements has also increased year by year. For a large amount of scientific and technological achievement data, how to efficiently and accurately recommend corresponding scientific and technological achievements to users, and how to efficiently and accurately recommend specific scientific and technological achievements to potential users are key issues that need to be solved. Summary of the invention

[0004] In order to solve some or all of the technical problems existing in the above-mentioned prior art, the present invention provides an intelligent recommendation method for scientific and technological achievements based on feature clustering and similarity calculation matching.

[0005] The technical solution of the present invention is as follows:

[0006] A method for intelligently recommending scientific and technological achievements based on feature clustering and similarity calculation matching is provided, the method comprising:

[0007] Acquire scientific and technological achievement data, extract information from each scientific and technological achievement, and obtain key data information and text semantic information corresponding to each scientific and technological achievement. The source of scientific and technological achievement data is internal data collection, or internal data collection and external data collection. Scientific and technological achievements include: at least one of technical reports, papers, and patent documents;

[0008] Matching the text semantic information corresponding to the scientific and technological achievements with the pre-determined field information to obtain the fields corresponding to the scientific and technological achievements, clustering the key data information corresponding to the scientific and technological achievements to obtain the clustering feature words corresponding to the scientific and technological achievements, and determining the field clustering feature words corresponding to each field according to the fields corresponding to the scientific and technological achievements and the clustering feature words corresponding to the scientific and technological achievements;

[0009] Acquire the user's function information and demand information, and construct user demand analysis information according to the field clustering feature words corresponding to each field, wherein the user demand analysis information includes: the field clustering feature words corresponding to the user;

[0010] Acquire the scientific and technological achievements of the user, classify each of the scientific and technological achievements of the user using a pre-trained classification model, obtain the field corresponding to each of the scientific and technological achievements of the user, and construct the analysis information of the field involved by the user, wherein the analysis information of the field involved by the user includes: a mapping relationship between the user and the field;

[0011] Based on the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information, determine the scientific and technological achievements recommended to the user.

[0012] In some optional implementations, the key data information includes: technical principles, application goals and conversion benefits.

[0013] In some optional implementations, the TextAbstractRank algorithm is used to extract a text summary from scientific and technological achievements, and the technical principles, application goals and conversion benefits are respectively extracted from the text summary.

[0014] In some optional implementations, the TextAbstractRank algorithm is used to extract a text abstract from a scientific and technological achievement, comprising the following steps:

[0015] Split scientific and technological achievements into multiple sentences;

[0016] Use nodes to represent sentences, use edges between nodes to represent the relationship between sentences, and construct a directed graph containing nodes and edges;

[0017] Assign initial weights to each node in the directed graph;

[0018] Using an iterative algorithm, the weight of each node is updated according to the relationship and weight between nodes until the preset convergence condition is reached;

[0019] According to the weight of the node, several nodes with the highest weight are selected as the text summary.

[0020] In some optional implementations, the text semantic information includes: named entities in scientific and technological achievements, key words in scientific and technological achievements, key sentences in scientific and technological achievements, syntactic analysis results of scientific and technological achievements, and time information in scientific and technological achievements.

[0021] In some optional implementations, the user demand analysis information is constructed in the following manner:

[0022] By marking the key information in the user's function information and demand information, function labels and demand labels are obtained;

[0023] Compare the similarity between each functional label and each demand label and the domain clustering feature word respectively, and determine the domain clustering feature word with the highest similarity to each label;

[0024] The domain clustering feature word with the highest similarity to each tag is used as the domain clustering feature word corresponding to the user to obtain user demand analysis information.

[0025] In some optional implementations, determining the scientific and technological achievements recommended to the user according to the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information includes the following steps:

[0026] When there is no browsing data of the user, the field and clustering feature words corresponding to the user's scientific and technological achievements are determined, and scientific and technological achievements in the same field and with the same clustering feature words are selected from other scientific and technological achievements and recommended to the user; or, when there is no browsing data of the user, similarity calculation is performed on the key data information corresponding to the user's scientific and technological achievements and the key data information corresponding to other scientific and technological achievements, and at least one scientific and technological achievement with the highest similarity to the user's scientific and technological achievements is selected and recommended to the user.

[0027] In some optional implementations, determining the scientific and technological achievements recommended to the user according to the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information includes the following steps:

[0028] When there is user browsing data, the user's browsing data within a specified time period is selected, the field type and cluster feature word information corresponding to the selected browsing data are determined, and scientific and technological achievements belonging to the determined field type range and having at least one cluster feature word identical to the determined cluster feature word are selected from other scientific and technological achievements to be recommended to the user; or, when there is user browsing data, the field type and cluster feature word information corresponding to the user's browsing data are determined, and scientific and technological achievements belonging to the determined field type range and having at least one cluster feature word identical to the determined cluster feature word are selected from other scientific and technological achievements to be recommended to the user.

[0029] In some optional implementations, determining the scientific and technological achievements recommended to the user based on the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information, and the user's field analysis information further includes the following steps:

[0030] A similarity calculation process is performed on the key data information corresponding to the user's scientific and technological achievements and the key data information corresponding to other scientific and technological achievements, and at least one scientific and technological achievement having the highest similarity to the user's scientific and technological achievements is selected and recommended to the user.

[0031] In some optional implementations, determining the scientific and technological achievements recommended to the user based on the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information, and the user's field analysis information further includes the following steps:

[0032] The field and field clustering feature words corresponding to the user are determined, and scientific and technological achievements belonging to the field corresponding to the user and having at least one clustering feature word identical to the determined field clustering feature word corresponding to the user are selected from other scientific and technological achievements and recommended to the user.

[0033] The main advantages of the technical solution of the present invention are as follows:

[0034] The scientific and technological achievements intelligent recommendation method based on feature clustering and similarity calculation matching of the present invention can efficiently and accurately extract corresponding scientific and technological achievements from a large amount of scientific and technological achievements data and recommend them to corresponding users, as well as efficiently and accurately recommend specific scientific and technological achievements to potential users to support users' scientific research. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The drawings described herein are used to provide a further understanding of the embodiments of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0036] Figure 1 The present invention is a flowchart of a method for intelligently recommending scientific and technological achievements based on feature clustering and similarity calculation matching according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0038] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or equipment.

[0039] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0040] The present invention discloses a method for intelligently recommending scientific and technological achievements based on feature clustering and similarity calculation matching, which can efficiently and accurately extract corresponding scientific and technological achievements from a large amount of scientific and technological achievement data and recommend them to corresponding users, and efficiently and accurately recommend specific scientific and technological achievements to potential users to support the scientific research of users. The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings.

[0041] Embodiment 1

[0042] refer to Figure 1 , Figure 1 The flowchart of a method for intelligently recommending scientific and technological achievements based on feature clustering and similarity calculation matching disclosed in an embodiment of the present invention. To solve the above technical problems, an embodiment of the present invention provides a method for intelligently recommending scientific and technological achievements based on feature clustering and similarity calculation matching, the method comprising the following steps 1-5:

[0043] Step 1: Obtain scientific and technological achievement data, extract information from each scientific and technological achievement, and obtain key data information and text semantic information corresponding to each scientific and technological achievement;

[0044] In the embodiment of the present invention, the source of scientific and technological achievements data may be internal data collection. Internal data is directly derived from an internal database or collected manually. Scientific and technological achievements include: at least one of technical reports, papers, and patent documents.

[0045] In the embodiment of the present invention, the source of scientific and technological achievement data can also be internal data collection and external data collection. Internal data is directly exported from an internal database or collected manually. External data is collected from a designated website manually or through some collection software and other tools, and the designated website is, for example, an academic website such as CNKI. Scientific and technological achievements include: at least one of technical reports, papers, and patent documents.

[0046] In the embodiment of the present invention, the key data information includes: technical principles, application goals and conversion benefits.

[0047] In the embodiment of the present invention, the text semantic information includes: named entities in scientific and technological achievements, key words in scientific and technological achievements, key sentences in scientific and technological achievements, syntactic analysis results of scientific and technological achievements, and time information in scientific and technological achievements. Among them, the syntactic analysis results include segmentation and grammatical relations between segmentations.

[0048] Step 2, matching the text semantic information corresponding to the scientific and technological achievements with the pre-determined field information to obtain the fields corresponding to the scientific and technological achievements, clustering the key data information corresponding to the scientific and technological achievements to obtain the clustering feature words corresponding to the scientific and technological achievements, and determining the field clustering feature words corresponding to each field according to the fields corresponding to the scientific and technological achievements and the clustering feature words corresponding to the scientific and technological achievements;

[0049] In the embodiment of the present invention, the specific number and specific type of fields are set according to actual needs. For example, the fields are set to ten, and the ten fields are: power and energy field, detection and identification field, functional materials and manufacturing field, navigation guidance and control field, artificial intelligence field, unmanned system field, information system and information security field, structural materials and manufacturing field, components and test instruments field and technology intersection field.

[0050] In the embodiment of the present invention, a matching algorithm is used to match text semantic information corresponding to the scientific and technological achievements with pre-determined field information to determine the closest field corresponding to the scientific and technological achievements.

[0051] In the embodiment of the present invention, the key data information of the scientific and technological achievements is clustered by using the k-means clustering algorithm to determine the clustering feature words corresponding to the scientific and technological achievements.

[0052] In the embodiment of the present invention, the domain clustering feature words are determined by the domain corresponding to the scientific and technological achievements and the clustering feature words corresponding to the scientific and technological achievements. Specifically, the domain clustering feature words include the clustering feature words corresponding to all scientific and technological achievements corresponding to the current domain.

[0053] Step 3: Obtain the user's functional information and demand information, and construct user demand analysis information based on the field clustering feature words corresponding to each field;

[0054] In the embodiment of the present invention, the user demand analysis information includes: domain clustering feature words corresponding to the user.

[0055] In the embodiment of the present invention, the user's functional information is determined according to the user's actual situation, for example, the user's functional information is the user's specific scientific research direction.

[0056] In an embodiment of the present invention, the user's demand information is determined according to the user's actual situation. For example, the user's demand information includes demand information in multiple dimensions such as technical requirements, application requirements, operational requirements, and capability requirements.

[0057] Step 4: Obtain the user's scientific and technological achievements, classify each of the user's scientific and technological achievements using the pre-trained classification model, obtain the field corresponding to each of the user's scientific and technological achievements, and construct the user's field analysis information;

[0058] In the embodiment of the present invention, the user-related domain analysis information includes: a mapping relationship between users and domains.

[0059] It should be noted that a user can be mapped to one domain or multiple domains.

[0060] Step 5, determining the scientific and technological achievements recommended to the user based on the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information.

[0061] In an embodiment of the present invention, based on the user's browsing data actually obtained, as well as the user's scientific and technological achievements, user demand analysis information and user-related field analysis information determined in the above steps 1 to 4, the scientific and technological achievements recommended to the user are determined from other scientific and technological achievements.

[0062] It should be noted that other scientific and technological achievements refer to scientific and technological achievements other than the users' scientific and technological achievements.

[0063] The intelligent recommendation method for scientific and technological achievements based on feature clustering and similarity calculation matching provided by the embodiment of the present invention extracts information from scientific and technological achievements, constructs user demand analysis information and user-involved field analysis information, and recommends scientific and technological achievements based on user browsing data, user scientific and technological achievements, user demand analysis information and user-involved field analysis information. It can efficiently and accurately recommend corresponding scientific and technological achievements to users, and efficiently and accurately recommend specific scientific and technological achievements to potential users.

[0064] Furthermore, in an optional implementation of the embodiment of the present invention, in the above step 1, the technical principles, application objectives and transformation benefits are extracted from scientific and technological achievements.

[0065] In an embodiment of the present invention, a machine learning method is used to extract a text summary from scientific and technological achievements, and then the technical principles, application goals and conversion benefits are respectively extracted from the text summary.

[0066] In the embodiment of the present invention, the TextAbstractRank algorithm is used to extract text abstracts from scientific and technological achievements.

[0067] Specifically, the TextAbstractRank algorithm is used to extract text abstracts from scientific and technological achievements, including the following steps 101 to 105:

[0068] Step 101, dividing the scientific and technological achievements into multiple sentences;

[0069] Step 102, using nodes to represent sentences, using edges between nodes to represent relationships between sentences, and constructing a directed graph including nodes and edges;

[0070] In the embodiment of the present invention, the relationship between sentences represents the similarity or connection strength between the sentences.

[0071] Step 103, assigning an initial weight to each node in the directed graph;

[0072] In the embodiment of the present invention, the initial weight is calculated according to the similarity between sentences.

[0073] Step 104, using an iterative algorithm to update the weight of each node according to the relationship and weight between the nodes until a preset convergence condition is reached;

[0074] In the embodiment of the present invention, the iterative algorithm is a PageRank algorithm.

[0075] In the embodiment of the present invention, the convergence condition is set according to actual needs. For example, the convergence condition is set as follows: the weight changes of all nodes before and after one iteration are less than a preset threshold.

[0076] It should be noted that the preset threshold is set according to actual needs.

[0077] Step 105, according to the weights of the nodes, select several nodes with the highest weights as text summaries;

[0078] In the embodiment of the present invention, all nodes are sorted according to the weights of the nodes, and multiple nodes with higher weights are selected as text summaries.

[0079] The number of selected nodes is set according to actual needs.

[0080] In the embodiment of the present invention, the above processing method can automatically extract the most important sentences from the scientific and technological achievements to form a text summary, and can obtain accurate and concise text summary information, and further obtain accurate technical principles, application goals and conversion benefit information.

[0081] Furthermore, in an optional implementation of the embodiment of the present invention, in the above step 1, a global pointer algorithm is used to identify and extract named entities from scientific and technological achievements.

[0082] Specifically, the global pointer algorithm is used to identify and extract named entities from scientific and technological achievements, including the following steps 111-112:

[0083] Step 111, extracting span features from scientific and technological achievements based on a preset span for representing a named entity;

[0084] In an embodiment of the present invention, the concept of span is used to represent entities, that is, an entity consists of a segment of continuous characters in a text. This representation method can avoid the impact of word segmentation errors that may occur in traditional word-based methods and can capture the boundaries of entities more accurately.

[0085] In the embodiment of the present invention, the span features include: part of speech, word vector representation, and character level features of the words in the span. These features are intended to capture the contextual information and linguistic features of the named entity.

[0086] Step 112, input the extracted span features into a pre-trained global pointer network to obtain a classification result corresponding to each span feature.

[0087] In the embodiment of the present invention, the classification result includes whether the span feature is a named entity or not a named entity.

[0088] In an embodiment of the present invention, the global pointer network is provided with an attention mechanism, which adaptively focuses on important contextual information by utilizing the attention mechanism and can take into account the global context, thereby more accurately determining whether the span feature is a named entity.

[0089] In the embodiment of the present invention, a cross entropy loss function is used as the loss function of the global pointer network during training.

[0090] In an embodiment of the present invention, when training a global pointer network, a plurality of training data including span feature sample information and its corresponding named entity category labels are obtained in advance, and based on a given loss function, the span feature sample information is used as the input of the global pointer network, and the named entity category labels corresponding to the span feature sample information are used as the output of the global pointer network to train the global pointer network.

[0091] In the embodiment of the present invention, the above processing method can effectively identify named entities in scientific and technological achievements, avoid problems such as word segmentation errors and insufficient local information in traditional methods, and improve the accuracy and efficiency of named entity recognition and extraction.

[0092] Furthermore, in an optional implementation of the embodiment of the present invention, in the above step 1, KeywordsRank algorithm is used to extract keywords from scientific and technological achievements.

[0093] Specifically, the KeywordsRank algorithm is used to extract keywords from scientific and technological achievements, including the following steps 121-125:

[0094] Step 121, dividing the scientific and technological achievements into multiple words;

[0095] Step 122, using nodes to represent words, using edges between nodes to represent the relationship between words, and constructing a directed graph including nodes and edges;

[0096] In the embodiment of the present invention, the relationship between words represents the co-occurrence relationship between the words. Specifically, when two words co-appear in scientific and technological achievements, there is an edge between the two words.

[0097] Step 123, assigning an initial weight to each node in the directed graph;

[0098] In the embodiment of the present invention, the initial weights of all nodes are set to the same value.

[0099] Step 124, using an iterative algorithm to update the weight of each node according to the relationship and weight between the nodes until a preset convergence condition is reached;

[0100] In the embodiment of the present invention, the iterative algorithm is a PageRank algorithm.

[0101] In the embodiment of the present invention, the convergence condition is set according to actual needs. For example, the convergence condition is set as follows: the weight changes of all nodes before and after one iteration are less than a preset threshold.

[0102] It should be noted that the preset threshold is set according to actual needs.

[0103] Step 125, according to the weights of the nodes, select several nodes with the highest weights as key words;

[0104] In the embodiment of the present invention, all nodes are sorted according to the weights of the nodes, and multiple nodes with higher weights are selected as key phrases.

[0105] The number of selected nodes is set according to actual needs.

[0106] In the embodiment of the present invention, the most important keywords can be automatically extracted from the scientific and technological achievements through the above processing method.

[0107] Furthermore, in an optional implementation of the embodiment of the present invention, in the above step 1, the TextAbstractRank algorithm is used to extract key sentences from scientific and technological achievements.

[0108] Specifically, the TextAbstractRank algorithm is used to extract key sentences from scientific and technological achievements, including the following steps 131-135:

[0109] Step 131, dividing the scientific and technological achievements into multiple sentences;

[0110] Step 132, using nodes to represent sentences, using edges between nodes to represent relationships between sentences, and constructing a directed graph including nodes and edges;

[0111] In the embodiment of the present invention, the relationship between sentences represents the similarity or connection strength between the sentences.

[0112] Step 133, assigning an initial weight to each node in the directed graph;

[0113] In the embodiment of the present invention, the initial weight is calculated according to the similarity between sentences.

[0114] Step 134, using an iterative algorithm to update the weight of each node according to the relationship and weight between the nodes until a preset convergence condition is reached;

[0115] In the embodiment of the present invention, the iterative algorithm is a PageRank algorithm.

[0116] In the embodiment of the present invention, the convergence condition is set according to actual needs. For example, the convergence condition is set as follows: the weight changes of all nodes before and after one iteration are less than a preset threshold.

[0117] It should be noted that the preset threshold is set according to actual needs.

[0118] Step 135, according to the weights of the nodes, select several nodes with the highest weights as key sentences;

[0119] In the embodiment of the present invention, all nodes are sorted according to the weights of the nodes, and multiple nodes with higher weights are selected as key sentences.

[0120] The number of selected nodes is set according to actual needs.

[0121] In the embodiment of the present invention, the most important key sentences can be automatically extracted from the scientific and technological achievements through the above processing method.

[0122] Furthermore, in an optional implementation of the embodiment of the present invention, in the above step 1, the LTP (Language Technology Platform) tool is used to perform syntactic analysis of the scientific and technological achievements to obtain syntactic analysis results. The LTP tool performs syntactic analysis of the text by using dependency syntactic analysis technology.

[0123] Specifically, the LTP tool is used to perform syntactic analysis of scientific and technological achievements, including the following steps 141 to 144:

[0124] Step 141: Segment the scientific and technological achievements and mark the part of speech for each word;

[0125] In the embodiment of the present invention, the scientific and technological achievements are segmented, each word in the scientific and technological achievements is separated, and the part of speech is marked for each word.

[0126] Step 142, constructing a syntactic analysis tree based on the word segmentation and part-of-speech tagging results, with each word as a node and the dependency relationship between words as an edge;

[0127] In the embodiment of the present invention, on the basis of determining the word segmentation and part-of-speech tagging, a syntactic analysis tree is constructed using dependency syntactic analysis technology. Dependency syntactic analysis is a method for representing sentence structure based on the dependency relationship between words in a sentence. In the syntactic analysis tree, each word is used as a node, and the dependency relationship between words is used as an edge to form a tree structure.

[0128] Step 143, extracting the grammatical relationship between the words according to the syntactic analysis tree and the dependency relationship between the words;

[0129] By analyzing the dependency relationship between words, the grammatical relationship between words can be extracted, such as subject-predicate relationship, verb-object relationship, etc. These grammatical relationships reflect the grammatical structure between words in a sentence and can provide important information for subsequent semantic analysis.

[0130] In the embodiment of the present invention, accurate syntactic analysis results can be obtained through the above processing method.

[0131] Furthermore, in an optional implementation of the embodiment of the present invention, in the above step 1, a Recognizers-Text algorithm is used to extract time information from scientific and technological achievements, and the time information is converted into a standard format.

[0132] Specifically, the Recognizers-Text algorithm is used to extract time information from scientific and technological achievements, and the time information is converted into a standard format, including the following steps 151-154:

[0133] Step 151, analyzing the scientific and technological achievements, identifying and extracting words or phrases that may represent time;

[0134] In the embodiment of the present invention, the words or phrases that may represent time refer to words or phrases containing at least one of “date”, “time”, “week” and “time period”.

[0135] Step 152, matching the words or phrases extracted by recognition based on predefined rules to obtain time expression content;

[0136] In the embodiment of the present invention, the predefined rules may include: regular expressions and grammar rules, which are used to identify and extract the specific content of the time expression.

[0137] Step 153, using a language model or statistical method to perform language understanding on the obtained time expression content to determine the accurate time meaning;

[0138] In the embodiment of the present invention, the language model or statistical method may adopt any existing language model or statistical method, and is not limited here. For example, the specific date, time, time zone and other information of the time expression may be inferred according to the context.

[0139] Step 154, converting the determined time meaning into a standardized time format.

[0140] In the embodiment of the present invention, the standardized time format is specifically set according to actual needs. For example, the date is converted to a unified date format, the time is converted to a 24-hour system, and the time zone information is ensured to be correct.

[0141] In the embodiment of the present invention, the above processing method can effectively identify and standardize the time expression in the text, provide the user with accurate and consistent time information, and improve the efficiency and accuracy of time processing.

[0142] In the embodiment of the present invention, when matching the text semantic information corresponding to the scientific and technological achievements with the predetermined field information to determine the closest field corresponding to the scientific and technological achievements, the existing matching algorithm is used, which will not be described in detail here.

[0143] In the embodiment of the present invention, when clustering the key data information of the scientific and technological achievements and determining the clustering feature words corresponding to the scientific and technological achievements, an existing clustering algorithm is used, which will not be described in detail here.

[0144] Further, in an optional implementation of the embodiment of the present invention, in the above step 3, constructing user demand analysis information includes the following steps 301-303:

[0145] Step 301, obtaining a function label and a requirement label by marking key information in the user's function information and requirement information;

[0146] In the embodiment of the present invention, a manual labeling method is used to label key information in the function information and the demand information to obtain a function label and a demand label.

[0147] Step 302, respectively compare the similarity between each functional label and each demand label and the domain clustering feature word, and determine the domain clustering feature word with the highest similarity to each label;

[0148] It should be noted that the domain clustering feature words corresponding to different tags may be the same.

[0149] Step 303: The domain clustering feature word with the highest similarity to each tag is used as the domain clustering feature word corresponding to the user to obtain user demand analysis information.

[0150] In the embodiment of the present invention, through the above processing method, user demand analysis information can be accurately constructed to improve the accuracy of subsequent achievement recommendation.

[0151] Furthermore, in an optional implementation of an embodiment of the present invention, in the above step 4, the classification model adopts a BERT model with a classification layer set at the output end.

[0152] In an embodiment of the present invention, when training a classification model, a plurality of training data including scientific and technological achievement sample information and its corresponding domain category labels are obtained in advance, the scientific and technological achievement sample information is used as the input of the classification model, and the domain category labels corresponding to the scientific and technological achievement sample information are used as the output of the classification model to train the classification model.

[0153] In an embodiment of the present invention, when training a classification model, a cross entropy loss function is used as a loss function during training.

[0154] In the embodiment of the present invention, in order to further improve the classification accuracy, when using the classification model to classify scientific and technological achievements, the scientific and technological achievements are first preprocessed to remove interference information such as irrelevant characters, punctuation marks, special symbols, etc. in the text.

[0155] In the embodiment of the present invention, through the above processing method, it is possible to accurately construct the user's field analysis information and improve the accuracy of subsequent achievement recommendation.

[0156] Further, in an optional implementation of the embodiment of the present invention, in the above step 5, determining the scientific and technological achievements recommended to the user according to the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information includes the following steps:

[0157] Step 501: when there is no browsing data of the user, determine the field and clustering feature words corresponding to the user's scientific and technological achievements, and select scientific and technological achievements in the same field and with the same clustering feature words from other scientific and technological achievements to recommend to the user; or, when there is no browsing data of the user, perform similarity calculation processing on the key data information corresponding to the user's scientific and technological achievements and the key data information corresponding to other scientific and technological achievements, and select at least one scientific and technological achievement with the highest similarity to the user's scientific and technological achievements to recommend to the user;

[0158] It should be noted that the user's browsing data refers to the scientific and technological achievement data that the user has browsed.

[0159] It should be noted that other scientific and technological achievements refer to scientific and technological achievements other than the users' scientific and technological achievements.

[0160] In the embodiment of the present invention, when performing similarity calculation processing, the technical principles, application goals and conversion benefit information in the key data information are processed. At the same time, since the key data information includes multi-dimensional text information such as technical principles, application goals and conversion benefits, when performing similarity calculation processing, the key data information is represented by text using machine learning methods, a representation vector is constructed, and similarity calculations are performed on vectors of different dimensions respectively, thereby obtaining similarity calculation results.

[0161] Step 502, when there is user browsing data, select the user's browsing data within a specified time period, determine the field type and cluster feature word information corresponding to the selected browsing data, select from other scientific and technological achievements that belong to the determined field type range and have at least one cluster feature word that is the same as the determined cluster feature word, and recommend them to the user; or, when there is user browsing data, determine the field type and cluster feature word information corresponding to the user's browsing data, select from other scientific and technological achievements that belong to the determined field type range and have at least one cluster feature word that is the same as the determined cluster feature word, and recommend them to the user.

[0162] It should be noted that the user's browsing data refers to the scientific and technological achievement data that the user has browsed.

[0163] It should be noted that other scientific and technological achievements refer to scientific and technological achievements other than the users' scientific and technological achievements.

[0164] In the embodiment of the present invention, by adopting the above-mentioned recommendation method, corresponding scientific and technological achievements can be efficiently and accurately extracted from a large amount of scientific and technological achievement data and recommended to corresponding users, and specific scientific and technological achievements can be efficiently and accurately recommended to potential users to support users' scientific research.

[0165] Furthermore, in an optional implementation of the embodiment of the present invention, in the above step 5, determining the scientific and technological achievements recommended to the user according to the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information, further includes the following steps:

[0166] Step 503 , performing similarity calculation processing on the key data information corresponding to the user's scientific and technological achievements and the key data information corresponding to other scientific and technological achievements, and selecting at least one scientific and technological achievement with the highest similarity to the user's scientific and technological achievements to recommend to the user.

[0167] It should be noted that the user's browsing data refers to the scientific and technological achievement data that the user has browsed.

[0168] It should be noted that other scientific and technological achievements refer to scientific and technological achievements other than the users' scientific and technological achievements.

[0169] In the embodiment of the present invention, when performing similarity calculation processing, the technical principles, application goals and conversion benefit information in the key data information are processed. At the same time, since the key data information includes multi-dimensional text information such as technical principles, application goals and conversion benefits, when performing similarity calculation processing, the key data information is represented by text using machine learning methods, a representation vector is constructed, and similarity calculations are performed on vectors of different dimensions respectively, thereby obtaining similarity calculation results.

[0170] In the embodiment of the present invention, by adopting the above-mentioned recommendation method, corresponding scientific and technological achievements can be efficiently and accurately extracted from a large amount of scientific and technological achievement data and recommended to corresponding users, and specific scientific and technological achievements can be efficiently and accurately recommended to potential users to support users' scientific research.

[0171] Furthermore, in an optional implementation of the embodiment of the present invention, in the above step 5, determining the scientific and technological achievements recommended to the user according to the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information, further includes the following steps:

[0172] Step 504, determine the field and field clustering feature words corresponding to the user, select scientific and technological achievements that belong to the field corresponding to the user and have at least one clustering feature word that is the same as the determined field clustering feature word corresponding to the user from other scientific and technological achievements, and recommend them to the user.

[0173] It should be noted that the user's browsing data refers to the scientific and technological achievement data that the user has browsed.

[0174] It should be noted that other scientific and technological achievements refer to scientific and technological achievements other than the users' scientific and technological achievements.

[0175] In the embodiment of the present invention, by adopting the above-mentioned recommendation method, corresponding scientific and technological achievements can be efficiently and accurately extracted from a large amount of scientific and technological achievement data and recommended to corresponding users, and specific scientific and technological achievements can be efficiently and accurately recommended to potential users to support users' scientific research.

[0176] It can be seen that the intelligent recommendation method for scientific and technological achievements provided by the embodiment of the present invention can efficiently and accurately extract corresponding scientific and technological achievements from a large amount of scientific and technological achievement data and recommend them to corresponding users, as well as efficiently and accurately recommend specific scientific and technological achievements to potential users to support users' scientific research.

[0177] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or partly contributed to the prior art in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, a magnetic disk storage, a magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0178] Finally, it should be noted that the method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching disclosed in the embodiment of the present invention is only a preferred embodiment of the present invention, which is only used to illustrate the technical scheme of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical schemes recorded in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical schemes from the spirit and scope of the technical schemes of the embodiments of the present invention.

Claims

1. A method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching, characterized in that: The method comprises: Acquire scientific and technological achievement data, extract information from each scientific and technological achievement, and obtain key data information and text semantic information corresponding to each scientific and technological achievement. The source of scientific and technological achievement data is internal data collection, or internal data collection and external data collection. Scientific and technological achievements include: at least one of technical reports, papers, and patent documents; Matching the text semantic information corresponding to the scientific and technological achievements with the pre-determined field information to obtain the fields corresponding to the scientific and technological achievements, clustering the key data information corresponding to the scientific and technological achievements to obtain the clustering feature words corresponding to the scientific and technological achievements, and determining the field clustering feature words corresponding to each field according to the fields corresponding to the scientific and technological achievements and the clustering feature words corresponding to the scientific and technological achievements; Acquire the user's function information and demand information, and construct user demand analysis information according to the field clustering feature words corresponding to each field, wherein the user demand analysis information includes: the field clustering feature words corresponding to the user; Acquire the scientific and technological achievements of the user, classify each of the scientific and technological achievements of the user using a pre-trained classification model, obtain the field corresponding to each of the scientific and technological achievements of the user, and construct the analysis information of the field involved by the user, wherein the analysis information of the field involved by the user includes: a mapping relationship between the user and the field; Based on the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information, determine the scientific and technological achievements recommended to the user.

2. The method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching according to claim 1, characterized in that: The key data information includes: technical principles, application objectives and conversion benefits.

3. The method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching according to claim 2 is characterized in that: The TextAbstractRank algorithm is used to extract text summaries from scientific and technological achievements, and the technical principles, application goals and conversion benefits are extracted from the text summaries respectively.

4. The method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching according to claim 3 is characterized in that: The TextAbstractRank algorithm is used to extract text abstracts from scientific and technological achievements, including the following steps: Split scientific and technological achievements into multiple sentences; Use nodes to represent sentences, use edges between nodes to represent the relationship between sentences, and construct a directed graph containing nodes and edges; Assign initial weights to each node in the directed graph; Using an iterative algorithm, the weight of each node is updated according to the relationship and weight between nodes until the preset convergence condition is reached; According to the weight of the node, several nodes with the highest weight are selected as the text summary.

5. The method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching according to claim 1, characterized in that: The text semantic information includes: named entities in scientific and technological achievements, key words in scientific and technological achievements, key sentences in scientific and technological achievements, syntactic analysis results of scientific and technological achievements and time information in scientific and technological achievements.

6. The method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching according to claim 1, characterized in that: The user demand analysis information is constructed in the following manner: By marking the key information in the user's function information and demand information, function labels and demand labels are obtained; Compare the similarity between each functional label and each demand label and the domain clustering feature word respectively, and determine the domain clustering feature word with the highest similarity to each label; The domain clustering feature word with the highest similarity to each tag is used as the domain clustering feature word corresponding to the user to obtain user demand analysis information.

7. The method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching according to any one of claims 1 to 6, characterized in that: According to the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information, the scientific and technological achievements recommended to the user are determined, including the following steps: When there is no browsing data of the user, the field and clustering feature words corresponding to the user's scientific and technological achievements are determined, and scientific and technological achievements in the same field and with the same clustering feature words are selected from other scientific and technological achievements and recommended to the user; or, when there is no browsing data of the user, similarity calculation is performed on the key data information corresponding to the user's scientific and technological achievements and the key data information corresponding to other scientific and technological achievements, and at least one scientific and technological achievement with the highest similarity to the user's scientific and technological achievements is selected and recommended to the user.

8. The method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching according to any one of claims 1 to 6, characterized in that: According to the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information, the scientific and technological achievements recommended to the user are determined, including the following steps: When there is user browsing data, the user's browsing data within a specified time period is selected, the field type and cluster feature word information corresponding to the selected browsing data are determined, and scientific and technological achievements belonging to the determined field type range and having at least one cluster feature word identical to the determined cluster feature word are selected from other scientific and technological achievements to be recommended to the user; or, when there is user browsing data, the field type and cluster feature word information corresponding to the user's browsing data are determined, and scientific and technological achievements belonging to the determined field type range and having at least one cluster feature word identical to the determined cluster feature word are selected from other scientific and technological achievements to be recommended to the user.

9. The method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching according to any one of claims 1 to 6, characterized in that: Determining the scientific and technological achievements recommended to the user based on the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information also includes the following steps: A similarity calculation process is performed on the key data information corresponding to the user's scientific and technological achievements and the key data information corresponding to other scientific and technological achievements, and at least one scientific and technological achievement having the highest similarity to the user's scientific and technological achievements is selected and recommended to the user.

10. The method for intelligent recommendation of scientific and technological achievements based on feature clustering and similarity calculation matching according to any one of claims 1 to 6, characterized in that: Determining the scientific and technological achievements recommended to the user based on the user's browsing data, the user's scientific and technological achievements, the user's demand analysis information and the user's field analysis information also includes the following steps: The field and field clustering feature words corresponding to the user are determined, and scientific and technological achievements belonging to the field corresponding to the user and having at least one clustering feature word identical to the determined field clustering feature word corresponding to the user are selected from other scientific and technological achievements and recommended to the user.

Images