Context- and sentiment-aware machine learning models for sentiment analysis

Download

index.pdf

Date

2023-1-24

Author

Deniz Kızılöz, Firdevsi Ayça

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

139
views

91
downloads

With the advances in information technologies, the amount of available data on web sources where people express their opinions increases continually. Sentiment analysis supports decision-makers in gaining insights from massive heaps of data. It has gained much attraction recently as it has proven to be a practical tool in a wide range of areas, including monitoring public opinion. Nevertheless, sentiment analysis research is still facing some challenges. One of the main challenges is the irrelevant and redundant features in the data. Such features not only increase the search space enormously but also disrupt the context awareness of the model. Another main challenge is the lack of domain-agnostic models for the sentiment analysis tasks as an existing model may not be the best fit for another domain in terms of context. Although deep learning models provide high-performance results, they require a massive amount of labeled data. However, obtaining a sufficient amount of labeled data is often impractical. In this thesis, we propose four models to remedy the aforementioned drawbacks. Our first model extracts the most informative features in the data for sentiment analysis. The second one constructs a context-refined word embedding model. The third model transfers the knowledge in pre-trained models to a new domain without the necessity of labeled data. The last one is a feature ensemble model that builds a pool of varying features for sentiment analysis. To verify the effectiveness of our models, we held extensive experiments on three benchmark datasets. Moreover, we introduced two novel datasets consisting of thousands of sentence and sentiment class pairs. Experiment results demonstrated that the proposed models yield performance improvements.

Subject Keywords

Natural language processing, Sentiment analysis, Machine learning

URI

https://hdl.handle.net/11511/102575

Collections

Graduate School of Natural and Applied Sciences, Thesis

Suggestions

OpenMETU
Core

Understanding IMF Decision-Making with Sentiment Analysis Deniz, Ayca; Angin, Merih; Angın, Pelin (2022-01-01) With the advances in information technologies, the amount of available data on web sources where people express their opinions increases continually. Sentiment analysis is one of the effective tools for decision-makers to gain insights from massive heaps of data. The field of International Organizations, which produces big data in the form of large documents, has significant potential to benefit from sentiment analysis in decision-making. In this paper, we evaluate the effectiveness of different sentiment a...
Sentiment analysis with recurrent neural networks on turkish reviews domain Rysbek, Darkhan; Uğur, Ömür; Department of Scientific Computing (2019) Easier access to computers, mobile devices, and availability of the Internet have given people the opportunity to use social media more frequently and with more convenience. Social media comes in many forms, including blogs, forums, business networks, review sites, and social networks. Therefore, social media generates massive sources of information in the shape of users‘ views, opinions, and arguments about various products, services, social events, and politics. By well-structuring and analysing this kind...
BIG DATA FOR INDUSTRY 4.0: A CONCEPTUAL FRAMEWORK Gökalp, Mert Onuralp; Kayabay, Kerem; Eren, Pekin Erhan; Koçyiğit, Altan (2016-12-17) Exponential growth in data volume originating from Internet of Things sources and information services drives the industry to develop new models and distributed tools to handle big data. In order to achieve strategic advantages, effective use of these tools and integrating results to their business processes are critical for enterprises. While there is an abundance of tools available in the market, they are underutilized by organizations due to their complexities. Deployment and usage of big data analysis t...
Real-Time Lexicon-Based Sentiment Analysis Experiments On Twitter With A Mild (More Information, Less Data) Arslan, Yusuf; Birtürk, Ayşe Nur; Djumabaev, Bekjan; Kucuk, Dilek (2017-12-14) Sentiment analysis of Twitter data is a well studied area, however, there is a need for exploring the effectiveness of real-time approaches on small data sets that only include popular and targeted tweets. In this paper, we have employed several sentiment analysis techniques by using dynamic dictionaries and models, and performed some experiments on limited but relevant datasets to understand the popularity of some terms and the opinion of users about them. The results of our experiments are promising.
Multi-modal learning with generalizable nonlinear dimensionality reduction Kaya, Semih; Vural, Elif; Department of Electrical and Electronics Engineering (2019) Thanks to significant advancements in information technologies, people can acquire various types of data from the universe. This data may include multiple features in different domains. Widespread machine learning methods benefit from distinctive features of data to reach desired outputs. Numerous studies demonstrate that machine learning algorithms that make use of multi-modal representations of data have more potential than methods with single modal structure. This potential comes from the mutual agreemen...

Citation Formats

F. A. Deniz Kızılöz, “Context- and sentiment-aware machine learning models for sentiment analysis,” Ph.D. - Doctoral Program, Middle East Technical University, 2023.