While the volume of available data has grown exponentially, most businesses struggle to derive value from it. Only a fraction of the potential value in terms of revenue and profit is captured. Vertical was founded at the beginning of 2017 with the vision to help companies create business opportunities out of the continuously growing volume of publicly available information. Our mission is to keep improving analytical capabilities to retrieve unstructured web data and analyse it at scale with the knowledge of a business expert. We enable businesses in evidence-based decision making in a time-saving manner and on a larger scale.
Our analysis is based on the same scalable, unified architecture:
We use different techniques for information retrieval and apply filters and quality checks to refine the input. A filter could be a client base, a competitors list, a risk- or investment portfolio, product codes or brand names. We search the web with use of different Application Programming Interfaces (API’s) to retrieve websites, pdf-reports, news articles and social media data.
The information retrieval process results in a database with pages of content. Depending on the content type we use, different (supervised) machine learning classifiers based on Watson’s natural language classifier, to label the content by it’s context. No document or website structure is the same. A human would navigate to a chapter, where the information is most likely to be found, our classifier does the same. This way there is no dependencies on webpage or document structure, and we analyse relevant content.
Next, we analyse specific content types per paragraph or sentence, with Watson’s human-like understanding; we program the business experts' knowledge and experience into Watson and return the content that is labeled as relevant by the business expert. For example a risk indicator or lead indicator.
It's an iterative self-learning proces, that will improve by the business experts input. Returned indicators are reviewed by clicking thumps up or thumps down. After sufficient training Watson recognizes similarities and ranks results different based on the feedback.
Curious what unstructured data analysis can do for you? Schedule an appointment with one of our consultants!