Analyzing A Deep Feature Extraction Method For App Descriptions

The advent of smartphones has created a new era of mobile devices,and mobile Applications carried on them have become one of the most important software in our daily life.The resulting App store not only provides convenience for users to download apps,but also brings fierce product competition.If an App wants to stand out in the App store,it is very important to understand the functions and detailed features of similar products.However,the App store contains a large number of App products,and it is difficult for product managers to fully understand all similar products when designing an App's functions.To solve this problem,this paper proposes the Deep Feature-based Competitor Analysis(DFCA)method,which is helpful for product managers to analyze competitors.App description is a description document for developers to introduce their products to users.In order to attract users to download,developers will introduce the characters of the product in the description document of the App in detail,so the description document contains rich information of App functions.At the same time,the development of Natural Language Processing technology provides technical support for mining the relevant information contained in App description documents,and the obtained relevant information can be used as a basis for comparing similar products to help product managers analyze competitors.Based on Natural Language Processing,data mining and other methods,combined with the characteristics of App software engineering,this paper studied App description documents,aiming to help product managers efficiently analyze the description documents of similar products,understand the status of similar products,and better analyze competitive products.The main research contents of this paper are as follows:(1)Designing a method to extract App features from App description documents.Feature is a sequence of words that exist in App-related documents to describe the functions of the software,and many existing App information mining works are carried out around feature extraction.In order to compare similar App functions more deeply,this paper defines a new App feature based on the existing work,which can contain the functions and details of the software at the same time.In addition,this paper uses Nature Language Processing tools to divide the App description document layer by layer,gradually identify the feature components of the App,and finally accurately extract such App features from the App description document.(2)Giving a method to mine competitive information among Apps.This paper further deals with the features extracted from the App description document,from which competitive information is mined and presented to the product manager.In order to find the similarity features,this paper designs an algorithm to calculate the similarity of features according to the actual situation.By using this algorithm,feature clustering can be achieved,and features describing similar App functions can be gathered together.After that,this paper further processed the details of similar features,so as to analyze the different features of similar functions,and help the product manager to conduct in-depth comparison.Above all,this paper analyzes App description and extracts App function-related information from description documents in the form of feature extraction.This paper presents a set of methods to analyze and compare the functions of similar apps automatically.It also provides references for the use of Natural Language Processing,text mining and other techniques in software engineering.