section{Introduction}Many new forms of communication have emerged in the last few decades, such as text messaging, and have become quite popular and important. These new forms of communication convey a wide range of information and are also commonly used to share feelings and opinions on different events and topics. We worked on the following task. The task is: egin{itemize}item Given a message, classify whether the message has a positive, negative, or neutral sentiment. For messages that convey both a positive and negative sentiment, you should choose the stronger one, whichever it is.end{itemize}pagebreaksection{Motivation}We often encounter many challenges when working with these informal texts such as tweets that when working with traditional texts as newswire data. Tweets are generally short and sharp: they should end with one or two sentences. And this makes the use of language very informal, along with a lot of new spellings, slang, new abbreviations like tc tor "take care", gr8 for "big" and so on. And along with all this we have hashtags that perform the equivalent task of tagging Twitter messages. Recently, the task of managing such challenges and automatically understanding the opinions conveyed by these tweets has become very popular and has become a subject of research. \An important aspect of tweets is that they contain highly structured data on different aspects of actual communication such as location, language, individuals, time, etc. Twitter tracks several relevant pieces of information in JSON format, and we can shape that information to our best use. This associated information is useful for a variety of purposes, including but not... middle of the paper... on tweets for training. Our method achieves good accuracy with relatively small data sizes.pagebreaksection{Future work} egin{itemize}item We have covered most of the features in our classification. Somewhat, we have not included the effect of the following features on classification accuracy. egin{itemize}item Take care of emotions conveyed by abbreviationssitem Analyze whether subsequent sentences in a tweet are more important. (For example giving more weight to a $2^{nd}$ line in a 2 line tweet.)end{itemize}item While it was clear from work done by others on the same problem that SVM tends to perform better than other classifiers, it would interesting to see how the hybrid of other classifiers (like the Naive Bayes classifier) ​​with SVM would perform. (In our work we tried a hybrid of lots of words with SVM which improved accuracy)end{itemize}