An approach to analyze suicidal tendency in blogs and tweets using Sentiment Analysis

Accepted: 20/Aug/2018, Online: 31/Aug/2018 Abstract— Sentiment Analysis(SA) or Opinion Mining is done to find the opinion of the users and customers to review and analyse their opinions on various products and services. It is one of the major tasks in business these days for knowing demands of customers. In this paper a sentiment analysis method for analyzing the suicidal tendencies in blogs and tweets is proposed. The proposed method uses concepts like Bag of words ,Part of Speech and Natural Language Processing for analzing the text.In life sometimes such a situation arises where the person finds himself trapped and suicide only seems to be the ultimate respite from all problems. Such a person may not share his mental condition verbally with anyone but may share it through tweets and messages. This paper is an approach to reach out a helping hand to such people by analyzing such tendencies in their messages. The proposed method uses Python Language module Textblob for performing different analysis task on the text.


I. INTRODUCTION
With advancement in technology and development of social media, information dispersion and content transmission online takes place in no time. Opinions on a new product can be gathered by stakeholders in fraction of seconds. There is a large growth of user-generated content in social networks, such as Twitter, Facebook and Whatsapp . Businesses are using this content for getting review for their products and services. This massive information is analyzed using varied technological tools from Artificial Intelligence (AI), Machine Learning, Natural Language Processing (NLP) and many others .
Internet has become a platform for sharing opinions and exchanging ideas. Social networking sites and apps like Whatsapp,Twitter, Facebook, Google+ have become very popular and fast mode of communicating views on topics and products, sharing ideas, pictures and messages.
Opinion Mining is commonly used for finding opinion of customers to know their view on products and services provided by different companies. It helps businesses in improving their products and services by conducting online reviews.
Some of the earlier works in the area of Sentiment Analysis used the Bag Of Words (BOW) model. In this model a document is mapped to a feature vector and which is then classified by machine learning techniques. This method is simple but leads to a lot of loss of information (Xia & Zong, 2010). Another feature that is popularly used for syntax analysis in Sentiment Analysis is the Part Of Speech Traditional approaches are time consuming as they rely on manual feature extraction. Some approaches using deep learning has shown excellent performance in NLP tasks which is a promising alternative to traditional methods. (Collobert et al., 2011). A study of different techniques for analyzing twitter data was also done which highlights all possible methods of Sentiment Anlysis. (Kharde & Sonawane,2016).
In this paper, we propose sentiment analysis approach for analyzing the suicidal tendencies in blogs and tweet messages. Suicidal tendency is a condition where a person is unable to find out a way and finds him to be trapped in a particular situation. Suicide only seems to be the ultimate respite from all problems. Such a person may not share his mental condition verbally with anyone but may share it through tweets and messages.
The rest of the paper is organized as follows. Section II describes the proposed Sentiment Analysis methods and the functions of Python that can be used for analysis. Experimental results are presented along with the methods. Finally, Section III draws conclusion and future scope of the work.

II. SENTIMENT ANALYSIS
Sentiment analysis uses Natural Language Processing (NLP) for analyzing attitudes, opinions, views and emotions from text, speech, tweets and database sources in the publicly available social networks .In Sentiment analysis the opinions in text are classified as positive, negative or neutral.
Sentiment Analysis includes many tasks such as sentiment extraction, classification, and summarization. It aims at analyzing sentiments, views, attitudes, opinions, emotions of people towards products, services, topics, organizations and also individuals.

Bag of Words (BOW) Analysis
In this a set of words pertaining to the area concerned is selected. The text to be checked is verified against the selected words to identify the extent to which the text matches the words and the area on the whole. Table 1 shows the classification of sentiments based on bag of words concept. All possible words that can be used by people who have a suicidal tendency are selected and classified as positive, negative and neutral.

WORD FREQUENCY ANALYSIS
There are two ways to get the frequency of a word or noun phrase using Textblob.The first is through the word_counts dictionary and the other is by use of words.count function.

POLARITY AND SUBJECTIVITY ANALYSIS
In Python the sentiment property of Textblob returns a named tuple of the form Sentiment(polarity, subjectivity). The polarity score is a float in the range -1.0 and 1.0 while the subjectivity is a float in the range 0.0 and 1.0 where 0.0 is very objective and 1.0 is very subjective.

III. CONCLUSION
In this paper a sensitive topic is taken into consideration for analysis that is -Suicide‖. Few ways of identifying the crucial words in the blogs and tweets is suggested. This will help in saving life of a person. A person with suicidal tendency needs help to recover from that situation. The help should be provided at the right time or else it will be of no use. Python language provides a Textblob module for Sentiment Analysis which is used for extracting and analyzing data from messages. Different ways of checking and analyzing the text is provided.