Doctoral Research
DESIGN AND DEVELOPMENT OF WEB SENTIMENT ANALYSIS ALGORITHMS TO ENHANCE THE ACCURACY USING SCORE
Doctor of Philosophy in Computer Science (as printed on the thesis title page), Kalinga University, Raipur.
Thesis record
Record details.
- Title
- DESIGN AND DEVELOPMENT OF WEB SENTIMENT ANALYSIS ALGORITHMS TO ENHANCE THE ACCURACY USING SCORE
- Author (as printed)
- CHABIR AHAMED A.
- Document type
- Doctoral thesis
- Degree
- Doctor of Philosophy in Computer Science — as printed on the thesis title page.
- Institution
- Kalinga University, Raipur, IndiaThe title page prints “Kalinga University Raipur”. The country is not printed on the title page.
- Research Guide
- Dr. Gyanendra Kumar Gupta
- Degree award year
- 2021 — per separate university award documentation.
- Submission date
- Not stated in the thesis.
- Research areas
- Web sentiment analysis; Opinion mining; Stock-price prediction; Web mining
Summary
Summary of the research.
Descriptive summary of the thesis body — pending author verification
The body of the thesis studies whether sentiment drawn from Twitter can help explain and predict the movement of company share prices. Tweets are collected through the Twitter API, classified by polarity with a lexical (positive and negative word) method, and combined into an overall sentiment score using a weighted average on a seven-point polarity scale.
The sentiment scores are compared with share prices taken from the NSE website, using Pearson correlation, multiple regression and Automatic Linear Modelling (ALM). The regression and ALM analyses predict a company's closing price from its opening price and its overall sentiment. The problem statement and conclusion also refer to an artificial neural network (ANN) approach. The conclusion states that the data were collected with R and analysed with Excel and SPSS.
Tweets were collected for 50 days, from 1 November to 20 December 2013; the thesis reports 114,500 tweets analysed, about 2,290 per day, and a correlation sample of 35 trading days. The results chapters name Bharathi Airtel, Titan Industries, Bosch and TCS. The conclusion refers to five application domains: telecommunication services, retail, automobiles, information technology and FMCG.
The thesis reports a strong relationship between sentiment score and share price for Bharathi Airtel, Titan Industries and TCS, and a weaker one for Bosch, which it left out of the later analysis. It reports that adding the sentiment score to the opening price lowers the information criterion compared with the opening price alone, and interprets this as improved prediction accuracy. Its stated limitations include that word polarity varies by domain and that the experiments used numerical datasets only.
The thesis lists ten objectives. Two of them concern creating balanced and imbalanced datasets from website reviews and analysing classifier performance on imbalanced data; this record reports no results for those two objectives.
Descriptive keywords
Web Sentiment Analysis; Opinion Mining; Twitter Sentiment; Sentiment Score; Stock Price Prediction; Correlation and Regression; Automatic Linear Modelling; Web Mining
These keywords describe the research for this record. The thesis does not print a keyword list.
How to cite
Suggested citation.
CHABIR AHAMED A. DESIGN AND DEVELOPMENT OF WEB SENTIMENT ANALYSIS ALGORITHMS TO ENHANCE THE ACCURACY USING SCORE. Doctoral thesis, Kalinga University, Raipur. Degree awarded 2021.
- 2021 is the documented degree-award year (per separate university award documentation).
- It is not a verified thesis publication or submission date.
- No DOI or institutional repository deposit is claimed.
- The complete thesis PDF remains private.
Author profiles
Profiles of the researcher.
Scope of this record
What this record documents.
This page records the bibliographic details of the doctoral thesis with a descriptive summary. The full text and the thesis’s own abstract are not reproduced on this website. The author name is shown as it is printed on the thesis title page.
Start a conversation