• Mr. Mithun M. Patil
• Mr. Vishwajit B. Gaikwad

Data available online in the form of text coming from various sources such as social media comments, product reviews, customer queries, search engine queries etc. is huge source of information, and extracting required knowledge out of it is valuable to the business. However processing and inferring meaningful information from this unstructured data is highly challenging task. First and foremost reason is data obtained from social networking comments, customer reviews are usually grammatically incorrect. Further it lacks sufficient statistical information to support many state-of-the-art approaches. Finally they are complex, ambiguous with misspelled words and are generated in an enormous volume, which further increases the difficulty to handle them. After studying multiple methods proposed recently in the field of text analysis it is observed that in order to infer the actual meaning of short text it is essential to have semantic knowledge. This work has proposed a prototype system that uses deep neural network for text processing. The proposed system has two phases namely Model building and Live testing. In first phase, KERAS sequential model is built and trained on YELP dataset of business reviews. In Live testing the user input query is processed in real time to derive its semantics and sentiment which can be either positive, negative or neutral. For getting semantic information the proposed method uses Simple Lesk algorithm and the sentiment is derive using KERAS model built in first phase. Proposed method is tested on Ebay customer review data and results are compared with some of the state-of-art methods namely TextBlob, VADAR analysis and SWN analysis. The results show our method is more effective than Sentiwordnet analysis, almost similar as VADAR analysis but lesser effective as compared to Textblob.

Text processing, Simple Lesk, KERAS Sequential model.

"Sentiment analysis of short text using KERAS", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.9, Issue 9, page no.b201-b205, September-2022, Available :http://www.jetir.org/papers/JETIR2209126.pdf

"Sentiment analysis of short text using KERAS", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.9, Issue 9, page no. ppb201-b205, September-2022, Available at : http://www.jetir.org/papers/JETIR2209126.pdf