| With the advent of the cloud computing era,the cost of creating,capturing and managing information has gradually decreased.The amount of data on the Internet is also showing explosive growth,more and more scientific and technological resources are uploaded to the network.Different from the news and social media data ubiquitous on the Internet,the main body of scientific and technological resources is composed of academic-style resources or entities such as papers,patents,authors,and research institutions.There is a rich relationship network between resources,from which a large amount of cutting-edge scientific and technological information can be mined.How to construct a complete and accurate representation of scientific and technological resources from structured and unstructured reports and texts in the network,and how to tap the potential value of scientific and technological resources is an urgent problem.The solution is to construct accurate portraits of scientific and technological resources in combination with knowledge-graph-related technologies.The main work of this thesis includes the following aspects:(1)This thesis proposes the acquisition and feature representation methods of scientific and technological resources.We use a Scrapy crawler to obtain scientific research-related papers and patent data on the Internet.A text feature representation method based on the Bert-FDP pre-training model is proposed,and the Bert model is fine-tuned on the natural language inference data set to obtain a more accurate and reasonable text vector representation,and realize the semantic feature representation of scientific and technological resource texts.(2)This thesis proposes a method for constructing accurate portraits of scientific and technological resources.This thesis proposes an entity extraction algorithm for scienti fic and technological subject words(PTMS)based on a pre-training model and multiple statistics,and extracts additional scientific and technological subject word entities from scientific and technological texts.A domain entity relation extraction algorithm(AHMC)based on attention mechanism and hierarchical multi-label classification is proposed to realize the extraction of association relations between scientific research achievements and hierarchical subject domains.Construct a knowledge map of scientific and technological resources entities and concepts to achieve a three-dimensional accurate portrait of scientific and technological resources(3)This thesis proposes the specific design of scientific and technological resources knowledge service components.Design and develop scientific and technological resource retrieval and query knowledge service components:to provide users with basic entity retrieval and relationship analysis functions;application domain query knowledge service components:for cross-field researchers to find different application scenarios of technology;subject area mining knowledge service components:display the parent-child fields of the subject area,as well as the representative scientific research results,scientific and technological keywords,journals and other scientific and technological resources under each field.Complete the design of the scientific and technological resources knowledge service component system.(4)This thesis implements the science and technology resource knowledge service component system.The system mainly includes the following three modules:the scientific and technological resource retrieval and query module,the application domain search knowledge service component module,and the subject domain mining knowledge service component module.The realization process of the functional modules of the system is expounded,and the main functions of the system are tested and verified. |
