1

Week 3 Homework

LAXMI RAVULA

Department of the Information Technology, University of the Cumberlands

ITS-632-B02: Intro to Data Mining (Second Bi-term)

Prof. Amit Karmaker

July 23, 2023

 2

Week 3 Homework

Answer 1

Hemmatian's (2019) study conducted a comprehensive survey of opinion mining

methods, addressing their challenges, and uncovering valuable insights. It acknowledged the

effectiveness of supervised approaches in accurately classifying comments but highlighted their

drawbacks of being slow and costly due to reliance on tagged training data. In contrast, the

research showcased a growing interest in semi-supervised methods, particularly for sentiment

analysis in microblogs like Twitter, where their low overhead and cost-effectiveness make them

favorable choices. The study revealed numerous organizations and companies widely employed

seven primary classification techniques and algorithms. Logistic regression, naive Bayes,

random forest, and decision trees stood out for their ease and efficiency in sentiment analysis

(Hemmatian & Sohrabi, 2017).

K-nearest and stochastic gradient descent were popular choices for specific classification

tasks. Furthermore, clustering and lexicon-based methods garnered increasing attention from

scholars, offering practical and up-to-date solutions for classifying comments in real-world

scenarios. Moreover, Hemmatian's research shed light on the landscape of opinion mining

methods, emphasizing the potential of semi-supervised approaches, the widespread adoption of

specific classification techniques, and the rising interest in clustering and lexicon-based methods.

These findings provide valuable insights for researchers and practitioners seeking effective

strategies for sentiment analysis across various contexts and datasets (Hemmatian & Sohrabi,

2017).
 3

In summary, the study emphasized the importance of considering the strengths and

challenges of different opinion mining methods. It highlighted the potential of semi-supervised

approaches, the prevalence of specific classification techniques, and the growing popularity of

clustering and lexicon-based methods within the sentiment analysis domain. These findings offer

valuable insights for researchers and practitioners seeking effective strategies for analyzing

sentiments in diverse datasets.

Answer 2

Opinion mining is a natural language processing (NLP) technique that involves

identifying and extracting opinions, sentiments, emotions, and attitudes expressed in text data,

such as user-generated content. The goal of opinion mining is to determine whether the

expressed views are positive, negative, or neutral and to understand the overall sentiment

polarity towards a particular subject, product, service, or topic (Pang & Lee, 2008). Opinion

mining employs various NLP techniques and machine learning algorithms to analyze sentiments

and opinions in text data.

In information retrieval, opinion mining is crucial in understanding and extracting valuable

insights from vast amounts of unstructured textual data on the internet and various online

platforms. Here's how opinion mining is used in information retrieval,

1. Product/Service Reviews

Opinion mining is widely used to analyze and summarize product or service reviews. It helps

businesses and consumers gain valuable insights into customers' overall sentiment, identifying

positive aspects (strengths) and negative aspects (weaknesses) of products or services (Pang &

Lee, 2008).

2. Brand Reputation Management

 4

Companies use opinion mining to monitor and assess their brand reputation by analyzing

online discussions, social media mentions, and customer feedback. This enables them to respond

to customer concerns promptly and improve customer satisfaction (Pang & Lee, 2008).

3. Market Research

Opinion mining aids in understanding customer preferences, expectations, and sentiments

related to specific products or industries. It gives businesses valuable market intelligence to

analyze informed decisions and develop effective marketing strategies (Pang & Lee, 2008).

4. Social Media Analysis

Social media platforms generate a heavy amount of textual data containing the sentiments

and opinions of users. Opinion mining analyzes and categorizes these sentiments, helping

businesses understand public perception and trends (Pang & Lee, 2008).

5. Lexicon-based Methods

These methods rely on sentiment lexicons or dictionaries containing words and their

associated sentiment polarities (positive, negative, or neutral). The view of a text is determined

based on the presence and frequency of sentimental words in the document (Pang & Lee, 2008).

6. Machine Learning

Supervised learning algorithms, such as SVM, Naive Bayes, and Decision Trees, can be

trained on labeled data (sentiment-labeled texts) to classify new texts as positive, negative, or

neutral (Pang & Lee, 2008).

7. Deep Learning

Deep neural networks, particularly Recurrent Neural Networks (RNNs) and Long Short-

Term Memory (LSTM) models, have shown significant success in sentiment analysis due to their
 5

ability to capture contextual information and long-range dependencies in text (Pang & Lee,

2008).

In summary, opinion mining is crucial in information retrieval across various domains. Its

ability to analyze sentiments and opinions from text data provides valuable insights for

businesses, researchers, and policymakers, facilitating better decision-making and enhancing

user experiences. As the volume of unstructured textual data grows, opinion mining will remain

essential in information retrieval.

Answer 3

Opinion mining is a part of natural language processing (NLP) that focuses on extracting and

analyzing sentiments, emotions, and subjective information from text data (Sun et al., 2017). It

involves understanding the underlying opinions expressed by individuals in various forms of

communication, such as reviews, social media posts, news articles, and customer feedback. Here

are some of the key concepts and techniques used in opinion mining,

1. Sentiment Lexicons

Sentiment lexicons or dictionaries are fundamental tools in opinion mining. They contain

words and phrases and their associated sentiment polarities (positive, negative, or neutral). These

lexicons serve as a basis for sentiment analysis, where the presence and frequency of sentiment

words in a text are used to determine the overall sentiment (Varathan et al., 2016).

2. Rule-Based Systems

Rule-based approaches use predefined linguistic rules and patterns to identify sentiment-

bearing phrases and infer the overall sentiment of the text. Linguists and domain experts design

these rules to handle specific cases or domain-specific language (Varathan et al., 2016).

3. Aspect-Based Sentiment Analysis

 6

Aspect-based sentiment analysis goes beyond the overall sentiment of a document and

focuses on identifying sensations towards specific aspects or features within the text. This

technique benefits product reviews, where users may express different opinions about various

product attributes (Varathan et al., 2016).

4. Domain Adaptation

Opinion mining often faces challenges when dealing with different domains or industries.

Domain adaptation techniques aim to transfer knowledge learned from one field (e.g., movie

reviews) to another part (e.g., product reviews) to improve sentiment analysis performance

(Varathan et al., 2016).

5. Emotion Analysis

In addition to sentiment analysis, emotion analysis aims to identify the underlying emotions

expressed in text. This goes beyond positive/negative sentiment and includes joy, sadness, anger,

fear, etc. (Varathan et al., 2016).

6. Unsupervised Learning

Unsupervised learning techniques like clustering are used when labeled data is scarce or

unavailable. Clustering methods group similar texts based on similarity, which can help identify

different sentiment clusters or themes in the data (Varathan et al., 2016).

7. Cross-Lingual Sentiment Analysis

Cross-lingual sentiment analysis deals with sentiment analysis in multiple languages. It

involves developing models to transfer knowledge learned from one language to another,

enabling sentiment analysis in multilingual data (Varathan et al., 2016).

8. Time-Series Analysis
 7

Time-series sentiment analysis involves studying sentiment patterns over time. It helps

understand how sentiments evolve and fluctuate over specific periods, such as during product

launches or political events (Varathan et al., 2016).

Transforming an organization's NLP (Natural Language Processing) framework holds

significant importance in today's data-driven world. Upgrading NLP capabilities offers several

compelling advantages (Sun et al., 2017). Firstly, it improves customer experience by

understanding sentiments and feedback, enabling organizations to tailor products and services

accordingly. Secondly, it enhances data analysis by processing vast amounts of unstructured

textual data from various sources, enabling data-driven decisions and identifying emerging

trends. Additionally, NLP improves information retrieval, automates customer support with

chatbots, and increases efficiency in handling queries. Leveraging NLP provides a competitive

advantage as organizations gain insights into market trends and competitor activities. Embracing

NLP technology future-proofs the organization and ensures relevance in the fast-paced digital

landscape, making it a crucial investment for any forward-thinking business (Sun et al., 2017).

In conclusion, opinion mining, or sentiment analysis, is pivotal in natural language

processing by extracting sentiments, emotions, and attitudes from text data. Supervised learning

approaches offer high accuracy but at the expense of being slow and costly, while semi-

supervised methods, particularly suitable for microblogs like Twitter, provide cost-effectiveness

and efficiency. Clustering and lexicon-based methods are gaining popularity for their practicality

in real-world sentiment classification. Transforming an organization's NLP framework is crucial

in today's data-driven landscape, enabling better customer experiences, insightful data analysis,

and informed decision-making.

 8

References

Hemmatian, F., & Sohrabi, M. K. (2017). A survey on classification techniques for opinion

mining and sentiment analysis. Artificial Intelligence Review, 52(3), 1495–1545.

https://doi.org/10.1007/s10462-017-9599-6

Pang, B., & Lee, L. (2008). Opinion Mining and Sentiment Analysis.

https://doi.org/10.1561/9781601981516

Sun, S., Luo, C., & Chen, J. (2017). A review of natural language processing techniques for

opinion mining systems. Information Fusion, 36, 10–25.

https://doi.org/10.1016/j.inffus.2016.10.004

Varathan, K. D., Giachanou, A., & Crestani, F. (2016). Comparative opinion mining: A Review.

Journal of the Association for Information Science and Technology, 68(4), 811–829.

https://doi.org/10.1002/asi.23716