Tag Extraction And Integration Based On Application Data

In recent years,with the rapid development of mobile Internet,mobile applications grow explosively.The problem that how to help users find the right application in the mass of applications needs to be resolved for download platforms.The shortcomings of traditional hot sorting method are very obvious-a large number of popular applications have to compete for limited exposure position;a large number of long tail applications didn't get opportunities of appearance—which is not conducive to the construction of application ecosystem.Tagging system is an effective way to solve the long tail of resource and help users manage and retrieve resource in web 2.0 era.There's little research on tagging system in APP field.Besides,application download platform,such as APP store,Google Play and so on,has no labeling function.However,with the explosive growth,it is imperative to solve the problem of long tail through building tagging system.This paper is devoted to exploring the new field and researching how to construct the APP field tagging system efficiently,automatically and accurately,which includes four aspects:· Data preprocessing:For APP data,the researcher carried out new word discovery and con-struction of stop word list.The experiments show that data preprocessing can significantly improve tag quality.· Tag extraction:Generalized common keywords extraction and tag recommendation algo-rithms.Besides,based on the characteristic of multi-dimensional text in APP data,three effective tag extraction algorithms(SemanticRank,RankScorel and RankScore2)are pro-posed to improve the efficiency of tag extraction.The algorithm is more suitable for APP dataset characteristics,and could get better label results.· Tag integration:Drawing on the idea of knowledge graph,the researcher introduced multiple external data sources to build synonyms and partial order integration tags,which contributes to a substantial improvement in APP coverage and recall rate,and solving the label clutter and irregularities.In addition,this paper proposed the APP tag tree method to manage the APP tags,which effectively solved the problem of multi-dimensional APP tags.· Tagging management system:This paper implemented the APP tagging management sys-tem,which integrates the result of data preprocessing,tag extraction,and tag integration.And the system provides a friendly interface and visual interface to facilitate label manage-ment personnel to maintain and manage labels.