Competitive Intelligence Mining Based On Knowledge Graph For Enterprises In Social Media Environment

With the rapid development of economic globalization and the massive popularity of the Internet,the number of enterprises is growing rapidly,and the competition and confrontation among enterprises is becoming more and more intense.The advent of the knowledge-based economy has affected the production and operation activities of enterprises in various aspects and has put forward higher requirements for the survival and development of enterprises.In addition to increasing the investment of capital and human resources,it is more crucial to use the correct and reasonable technical means and formulate strategic plans scientifically to enhance the core competitiveness of enterprises.Competitive intelligence is the key method and important basis for analyzing competitors,predicting market environment,grasping market trends and making strategic decisions.The acquisition and collection of competitive intelligence,to a certain extent,determines the quality and effectiveness of subsequent competitive intelligence analysis.Generally speaking,competitive intelligence is often difficult to obtain,and social media platforms contain a huge amount of data,which is an important reference value as a natural data source for competitive intelligence.However,the open and huge data puts forward strict requirements on technical means,and the large amount of unstructured or semi-structured data raises the complexity of competitive intelligence collection to a new level.In order to solve such problems,knowledge graphs are born.Although the development of open knowledge graph has almost matured,there is still much room for the development of vertical domain knowledge graphs.In this paper,we will take knowledge graph technology as the core,enterprise competitive intelligence as the guide,and social media platform as the data source,and build a system process to serve enterprises with decision support and strategy making.The main work of this paper is as follows.(1)For the data acquisition of social media platforms,microblog and We Chat are proposed as the main data sources.Considering that the Web information is huge and complicated,two corresponding crawler solutions are proposed based on keyword search,distinguishing the similarities and differences of platforms,to obtain a large amount of social media platform data quickly and efficiently,as a basis for subsequent competitive intelligence mining work,and to provide information source use cases.(2)The identification and extraction of the acquired unstructured or semistructured data are realized.Using natural language processing techniques,a series of pre-processing work such as word separation and annotation are performed on the acquired data.Meanwhile,entity alignment is performed in combination with open knowledge graph set,which can reduce the workload to a certain extent.For the problem of entity extraction,a model based on BiLSTM-CRF is proposed to identify named entities for text contents,and after the entities are identified,the extraction of relationships based on predefined entity relationships is proposed using BiLSTMAttention model.After completing the above work,the entities and relations are linked to make the knowledge graph base triad.And we use TF-IDF algorithm and LDA topic model for document processing to quickly obtain document topics,and subsequently combine with K-means algorithm for topic clustering of documents,so as to build on the knowledge graph basis for extracting and clustering documents.(3)Building on the above work,the open source graph database Neo4 j will be used to construct and store the knowledge graph,and the advantages and disadvantages of the graph database compared with other databases are elaborated.The recommended process of building knowledge graphs using Neo4 j is collated and summarized,and the visualization interface of some knowledge graphs is shown.At the same time,problems are proposed in combination with hypothetical samples,and problems are solved using the completed constructed knowledge graph applications.The experiments show that the method proposed in this paper can effectively build knowledge graph applications based on the data obtained from social media platforms,and to a certain extent,assist enterprises in competitive intelligence mining,thus providing support for their strategic decisions.Therefore,this paper has certain significance for knowledge graph construction and enterprise competitive intelligence mining.