Research On Computer Blog Recommendation Algorithm Based On Skill And Interest Network

Online technical exchange communities are important platforms for technology enthusiasts or practitioners to exchange and share computer and internet technology.Users in the technology community browse blogs,bookmark blogs,comment on blogs,like blogs,and blog every day.Analyzing the changes in user interests and skills from historical data generated by users will help the technology community to provide users with more accurate and personalized services,which is of great benefit to users and the technology community.Traditional recommendation algorithms often focus on the“user-item”associations,and do not consider the chronological order of the user-item relationship.Many time series models are studied at the level of user interest.However,users in the technology community not only have characteristics of interest but also skills.Therefore,considering the interests and skills of users in the recommendation has become a direction worth exploring.Most articles in the IT technology community are in the computer field.The com-puter field is developing and updating rapidly.Therefore,it's common for some new technical words to appear.Some special nouns have specific shorthand expressions in the community.It is not easy to obtain the articles directly using the word segmenta-tion tool.The computer field requires auxiliary word segmentation Noun dictionary to improve the accuracy of word segmentation.Using the word segmentation tool directly is not easy to accurately divide the article into words.Computer domain proper noun dictionary to assist word segmentation to improve the accuracy of word segmentation.In order to save the problem of new words constantly appearing in the computer field,this paper designs and implements an algorithm for new word discovery in the com-puter field to expand the dictionary of proper nouns in the computer field.In addition,the artilce in the technical community is Markdown format.Some symbols that origi-nally belong to stop words have become Markdown's grammatical marks in the article to enhance the semantic representation of words.Aiming at this feature of Markdown text,this paper proposes the concept of semantic blocks and designs and implements a Markdown feature to improve the accuracy of unsupervised keyword extraction.The Markdown feature is used to improve the TF-IDF algorithm and TextRank algorithm.Based on these two algorithms,this paper designes and implementes an unsupervised Markdown feature-aware keywords extraction algorithm(MDKE)to extract the key-words of articles in Markdown format.Considering that the users in the technical com-munity are both the author and reader of the article,the author reflects the output of user skills,and the reader reflects the user's interest orientation.This paper considers the articles that the user likes as the user's interests and the articles that the user post as the user's skill.Obvious chronological order,regard user like behavior as time series and use LSTM model to extract user's interest characteristics,and similarly treat user's post behavior to extract user's skill characteristics.Static characteristics such as whether the user is involved in the work and whether he is a domain author will also affect the recommendation result.This paper uses MLP models to extract higher-level representa-tions of user static features.Finally,this paper proposes a user interest and skill aware computer blog recommendation model(SKAIN)based on user static characteristics,user interest characteristics and user skill characteristicsthe,while taking into account user interests and skills.This paper designs experiments on the real data set of Jujin.Firstly,the MDKE algorithm is compared with the mainstream keyword extraction algorithm to verify the effectiveness of the MDKE algorithm in extracting keywords from Markdown format text.Secondly,this paper designs a comparative experiment for the SKAIN model to further verify the combination of user skills The advantages of the SKAIN model of fea-tures and interest features in computer blog recommendation.The research in this paper broadens new ideas for unsupervised keyword extraction and technical community ar-ticle recommendation.