| Natural language processing(NLP)is a challenging research in computer science to information management,text mining,and enabling computers to obtain meaning from human language processing in text-documents.NLP techniques have provided golden opportunities for extracting information and classifying documentation in text-documents.Topic modeling is one of the most powerful techniques in natural language processing for text mining,latent data discovery,and finding relationships among data and text documents.NLP based on topic model have attracted much attention for researchers from various subjects such as software engineering,media science,medical and linguistic science,etc.Recently,topic modelling and recommending highlight topics on digital humanities,Community Question and Answer(CQA)websites and online forums are challenging research in NLP area.Existing researches mostly focus on sentiment/polarity analysis from data and can be obtained for positive,neutral,and negative sentiments,which lack the use of semantically technique.So we can conduct further researches to overcome these drawbacks.Therefore,we cover both sentiment and semantic aspects with collaboration of hybrid frameworks based on unsupervised and NLP techniques,which can be practical for recommending meaningful-topics of online context.However,one of the main goals of this dissertation is investigating the novel capability of NLP methods to knowledge discovery and meaningful topic recommendation in different areas such as social media,scholarly community,healthcare science,and software engineering.Generally speaking,this dissertation has introduced five works,two new dataset and proposed three models based on the mentioned aspects.The main researches of this dissertation include:·A semantic model based on a novel application of NLP to discover the trends of the topics and find relationship between LDA topics and paper features.The semantic framework can retrieval and detect meaningful latent topics for recommending interesting research field of scholars to director conferences.·A collaborative framework based on LDA topic model and a random forest to retrieval latent topics on healthcare data from an online forum.According to our experiment on our healthcare data,we found that this model can be practical for both researchers and medical professionals to understand and recommending significant concerns/problems from patients in an online health community.·A novel hybrid framework is proposed by combining semantic,sentiment and fuzzy approaches,which can assist in satisfying viewers’ experience and improve recommendation systems of movies to the related groups.However,discovering and recommending meaningful-latent-topics of YouTube users comments in social media is another interesting and challenging area of natural language processing.·A hybrid-fuzzy framework based on the LDA topic model and fuzzy rules for question topic mining and recommending highlight latent topics in a community questionanswering forum of developer community.However,our model based on NLP techniques of Stack Overflow is very beneficial for pattern discovery and behaviour analysis in programming knowledge.·A systematic framework proposed based on a deep learning and NLP models that is capable of extracting meaningful topics from COVID-19-related comments in social media.In faxt,we designed a two-layers LSTM for sentiment classification,which produces better results compared with several other well-known.Moreover,our findings shed light on the importance of using public opinions and suitable computational techniques to understand issues surrounding COVID-19 and to guide related decision-making. |
