GetSentiment

We are an early stage start-up specializing in developing text analytics solutions for customer feedback analysis. We help market researchers, customer experience specialists and media analysts to extract from text valuable insights about customer opinions, preferences, dislikes, and intentions with respect to their brands. At the moment many companies possess lots of customer feedback data, a lot of which is in free-text format. The data can be feedback submitted through the company website, sent by email, posted on the company's social media pages, it can be transcriptions from the company's call centre, or a log of customers' chat sessions with the technical support. The data can also be "unsolicited feedback", such as any mentions of brands on news websites and on social media. The volumes of text-based feedback are such that it is unfeasible to have humans read and interpret them, so we help our clients to automate this process, extracting relevant information from text and making it amenable to statistical analysis. We have developed proprietary text analytics software which can be integrated into larger software projects via an API, provide services on customizing the software to specific uses cases, and develop bespoke integrated solutions that include data collection, processing, data visualization and reporting. There exist a large number of tools, both commercial and free/open-source ones, that perform text processing tasks such as sentiment analysis, keyword extraction, and text classification. We are one of the few providers of tools for aspect-based sentiment analysis, which not only assigns a sentiment score and category labels to the overall document, but works on the phrase level and identifies multiple categories of issues or "aspects" in a document as well as sentiment associated with each category. For example, a piece of customer feedback might include a sentence such as "I liked the service, but the prices are way too high." The positive comment is counterbalanced by the negative one, so the overall score may suggest neutrality. Instead, we detect that there two issues being commented on - service and prices, and that the former one is associated with positive sentiment, while the latter - with negative. The accuracy of text analysis depends a lot on how well the NLP component is customized to the specific industry that the texts come from, and sometimes to specific uses cases. So in addition to the API, we have developed a self-service interface where users themselves can customize the NLP analyzers to their needs by uploading custom lexical resources. We also offer services on developing such customizations for our clients.