Abstract
This paper presents an approach for company's reputation analysis using data mining techniques. It obtains knowledge from huge data written about these companies and available publicly on the internet. It is done by extracting data from social media, such as twitter, containing relevant company's mentions. Then, data is then injected into the first layer where it is classified into a positive and negative classes using machine learning, specifically artificial neural network. It takes tweets after preprocessing as an input, then outputs the sentiment of each tweet. The result will be the general reputation and perception of the company for a given timeframe. To further understand these results, the analyzed data is then transferred to a second layer of analysis where consumers to identify products, services, and announcements that lead to the positive of negative perception. Using Term Frequency (TF), this will result to ranked list of most mentioned words in each of the negative and positive classes. This will be valuable for companies to identify points of weakness and strength, advertisement impressions, and strategic decisions impact.