Often times we want to know what people think about something. For example, one of our clients wants to tell surgeons what other surgeons think about a particular aspect of a surgery they performed. Another client of ours wants to know if a potential lead for their customer's product is positively inclined toward a follow up sales call. We have yet another client of ours that wants to know not only how patients feel about their doctor, but how they feel about the clinic's environment, or the communication style of the physician, or some other aspect of their experience. In all of these cases, our clients have a lot of unstructured data that contain these hidden gems of sentiment. For example, their data might include natural language text like online reviews, or it might originate in audio recordings or transcriptions.
Generic off the shelf sentiment solutions typically offer simplistic assessments like positive or negative classifications, but rarely do they offer the sentiment about something, like the communication style of a physician, or the cheerfulness of the staff. In addition, generic solutions are trained on more general text, so if the speaking or writing style of your customers is unique or different, these generic solutions might not perform well. At Xyonix, our document parsers (machine text readers) are very easy for us to customize so they can very accurately read your documents, regardless of how cryptic or nuanced the language of your users may be.