Twitter Sentiment in Data Streams with Perceptron

Abstract

With the huge increase in popularity of Twitter in recent years, the ability to draw information regarding public sentiment from Twitter data has become an area of immense interest. Numerous methods of determining the sentiment of tweets, both in general and in regard to a specific topic, have been developed, however most of these functions are in a batch learning environment where instances may be passed over multiple times. Since Twitter data in real world situations are far similar to a stream environment, we proposed several algorithms which classify the sentiment of tweets in a data stream. We were able to determine whether a tweet was subjective or objective with an error rate as low as 0.24 and an F-score as high as 0.85. For the determination of positive or negative sentiment in subjective tweets, an error rate as low as 0.23 and an F-score as high as 0.78 were achieved.

Sentiment Analysis on the Social Networks Using Stream Algorithms

“…Our MBW achieved the highest accuracy of 87.5% with STS_Gold on 5 grams representation and only 73.3% on 3 grams and 73.6% on 5 grams for Sanders and Sentiment Strength respectively. Our results for Sanders are close to the accuracy of [9] for their 5 grams on Sanders. MBW is capable of performing feature selection dynamically on a date stream, instead of performing a batch feature selection beforehand, such as [9] does.…”

“…By incorporating dynamic feature selection in our MBW we achieved an accuracy of 73.3% while [9] achieved an accuracy of 77% using 5 grams with manual feature selection. In terms of a data stream, it is important to perform dynamic feature selection due to the changing importance of features with new incoming data.…”

“…Most closely related to the research done in this paper is the work of Aston, Liddle, and Hu [9] on Twitter sentiment classification in data stream. The Perceptron algorithm and its voted version with static feature selection were used to predict sentiment in a data stream environment for large amounts of Twitter data.…”

“…The work of [9] performed feature selection on a static batch and then performed sentiment analysis on this reduced feature set. This approach has downfalls as unknown data coming in over time will change the importance of features.…”

“…Thus we performed online feature selection to discover the top features of each gram representation [14]. Unlike static feature selection methods, which find the top features once in a batch learning environment [9], we performed feature selection dynamically, with a newly calculated set of top features at each time interval of 100 sequential posts. At first we removed all but the top features when performing classification but found that this resulted in a considerable reduction in accuracy.…”