Fine-Grained Sentiment Analysis Of Homestay Comments Based On Text Mining

The rapid development of Internet technology,information technology and economy has provided a good platform for the development of the shared economy.Sharing mode is spread in all aspects of life,and people's lives have changed with each passing day.The emergence of shared space mode brings diversified choices to people's travel and accommodation.Homestay,a new type of shared accommodation product,is very popular among traveling users because of its comfortable and free experience.In view of the history of sharing bicycles from prosperity to recession,the residential industry should pay attention to and reflect on its own development status and existing problems,and move forward with the goal of sustainable development.Therefore,using text mining technology to analyze the current situation of homestay is very meaningful.At present,online hotel reservation websites can realize online display of offline housing source information,which is convenient for consumers to browse,book and feedback online.Therefore,starting from the comments on the online hotel reservation website of the homestay,summarize the evaluation system of the characteristics of the homestay industry,excavate the users' emotional tendencies towards the characteristics of homestay,and provide hotel and homestay operators with intuitive user evaluation information,thus providing them with operational reference value.The following three parts are the main research work of this paper:(1)Emotional evaluation unit was extracted.First,we grab the comment data and preprocess it.An anti-crawler program based on Python is designed to crawl home reviews from Ctrip website.Customized Hotel domain proper noun dictionary for Jieba segmentation,to promote the hotel domain features to be fully segmented.A custom deactivated dictionary is used to filter meaningless words.The part-of-speech tagging is carried out with LTP tool,and the meaningless parts of speech and words are removed.Then the emotional evaluation unit is defined as the sequence of attribute features,emotional features and emotional degree features.Seven rules of part-of-speech sequence are proposed to extract the emotional evaluation unit.Special negative and affirmative sentences are taken into account,and words labeled as idioms are added as emotional tendency words.Finally,the semantic similarity with frequent features is used to filter the wrong emotional evaluation units.It enriches the existing rules of part-of-speech sequence and improves the accuracy of the emotional evaluation unit.(2)Establishing the evaluation system of residential characteristics.Based on the existing hotel feature classification research,expand other attribute feature classification of residential accommodation.Therefore,based on the extracted feature set,the classificationof residential features is studied in two steps.The first step is to pre-classify the hotel characteristics according to the hotel characteristics based on the existing hotel characteristics classification research.In the second step,hierarchical clustering algorithm is applied to classify the non-classified residential characteristics.Python's SciPy library is used to implement hierarchical clustering analysis of features.A method of computing the similarity of words based on dictionary and word vector is proposed to calculate the distance between individual samples in hierarchical clustering analysis,which provides high-quality input parameters for clustering analysis.The clustering results show that the proposed method is effective.Finally,the results of the first and second classification of residential characteristics are summarized and summarized,and the evaluation system of residential characteristics is formed.(3)Fine-grained emotional analysis of residential characteristics.Based on frequent features and feature evaluation system,fine-grained emotional value calculation is carried out by using emotional dictionary.Five characteristics and emotional tendencies of residential accommodation,emotional tendencies and overall situation of each feature category of residential accommodation,which are most concerned by users,are excavated.