Information extraction with two-layered ODNN and semantic analysis for opinion mining.

User textual reviews have become a significant data source for enhancing the efficacy of recommendation systems as e-commerce website use has increased. These evaluations feature nuanced customer opinions that typically reflect their product preferences. On the other hand, the majority of conventional recommender systems (RS) disregard such user feedback and hence fall short of accurately capturing users' unique product thoughts. The majority of these techniques rely on handcrafted and rule-based approaches, which are known to be labor-intensive and time-consuming. Although other approaches try to employ fine-grained user opinions to somewhat enhance the accuracy of recommendation systems. Their applicability is therefore restricted. To address the aforementioned issues, this research develops an opinion-mining recommendation system that combines Information Extraction with Two-Layered ODNN and Semantic Analysis. The first layer uses NB and MSGM to forecast the aspects specified in the feedback, while the second layer uses ODNN to specify the opinion orientation as well as sentiment analysis. First and foremost, data pre-processing is critical before entering data into models, as it aids in reaching high accuracy. Special methods were necessary for textual data, including tokenization, stop word deletion, and user name removal, all of which were used in this study. Second, the first layer may focus on the critical context words to the given aspect terms in sentences utilizing NB, MSGM, and Enthalpy measure. Last but not least, Second Layer Opinion The outcome of sentiment categorization aspect polarities opinion orientation as positive, negative, and neutral is predicted using ODNN. In terms of information, the results of the experiments suggest that our proposed strategy outperforms current methods in terms of accuracy, precision, and recall for the Amazon product review at 95.29%, 96.67%, and 96.18%. As well as yelp dataset analysis of accuracy, precision, and recall are 92.13%, 90.45%, and 96.04%. [ABSTRACT FROM AUTHOR]