Recommended Real Estate Agents Based On Sentiment Strength Of Reviews And Collaborative Filtering

With the implementation of the “Internet plus” policy,the transformation speed of real estate and other traditional industries has accelerated,the number of real estate websites has increased,the rating and review data from these websites are in an explosive growth state.Massive data results in information overload and also leads to recommendation research.Compared with coarse-grained user ratings,reviews can reflect the real feelings of users more accurately.Sentiment analysis based on user reviews can effectively extend the original ratings of users,which provide the reference for other users and also provide the important reference for enterprises to formulate development strategies,improve service quality and enhance friendliness of users.At present,most real estate websites only provide the sorting function based on the historical ratings and other indexes of agents.The recommendation function is not implemented.Although the sorting function can save users' time and energy,it doesn't consider the extension effect of agents' historical reviews on ratings.In addition,sorting by single index is unilateral.In view of the above problems,this thesis mainly researches on recommended real estate agents based on sentiment analysis and collaborative filtering,which aims to make personalized recommendations to users based on the sentiment information of reviews and agents' other indexes.First of all,since real estate agent reviews often contain information unrelated to agents' service and most existing sentiment analysis algorithms do not consider the strengths of sentiment words,applying the existing sentiment analysis algorithms to real estate agent reviews will result in low accuracy.Sentiment analysis algorithm based on syntactic rules and sentiment strengths is proposed.Sentiment dictionary sets are generated from both reviews data set and open-source sentiment dictionaries.Then syntactic rules are proposed based on real estate agents' reviews.Both sentiment dictionary sets and syntactic rules are used to analyze agents' reviews and finally we can get the tendency of each agent review.Secondly,the result of sentiment analysis and agents' other indexes are used to extend agents' original scores and the final recommendation scores are obtained as the primary recommendation reference.Then the KNN algorithm and the improved collaborative filtering recommendation algorithm are used to generate agents neighbor sets,which are used as the secondary recommendation reference.Finally,the TOP-N recommendation list for users is generated from the above two recommendation references.Agent reviews data set and agent other indexes data set crawled from the well-known real estate website are used to verify the proposed algorithm.The result shows that the proposed algorithm can effectively improve the accuracy of sentiment analysis by integrating syntactic rules and the strength of sentiment word in agent reviews;extending agents' original rating by the result of sentiment analysis and agents' other indexes and improving collaborative filtering recommendation algorithm is reasonable and effective in the field of real estate agent recommendation.