Cognitive Analytics is a field of research that focusses on discovering hidden patterns within data based on existing knowledge.
It combines the best practices of Machine Learning and Natural Language Processing to make sense of untapped data sources. This gives you the opportunity to analyse great amounts of unstructured data, for example the transcripts of a video archive. Big Data research means that you try to rapidly produce insights from structured data. Cognitive Analytics does the same but for unstructured data.
But what’s so special about this?
Well, it’s hard for computers to recognise patterns in unstructured data. However, humans are very good at filtering all kinds of information and make decisions based on this information. For example, if you speak to someone in a foreign language and you have a hard time understanding what he or she means, you’ll focus on body language, tone of voice and all kinds of other signs to search for answers. All these bits of information will help you understand what that person means. It’s very hard for computers to imitate this process, but it’s essentially what you try to do with Cognitive Analytics.
How to use Cognitive Analytics?
There are a lot of ways to use Cognitive Analytics. Today we’ll highlight two tools that might be useful to analyse transcripts of audio visual content.
Automatic summarization is part of natural language processing. It works by calculating the word frequency for an entire document. The most common words are stored and sorted, and each sentence of the document is scored by these words. Based on this calculation the top sentences will form the final summary. This algorithm can be used to understand big amounts of text very fast. You can imagine that these solutions can help students, analysts, researchers or basically anyone who spends a lot of time consuming big amounts of text to save time. At Zoom Media we use this algorithm to help our customers get the most value out of their data and make it ‘searchable’ for everybody within their company.
The amount of data an average company is collecting is growing every day. As we gather more information, it becomes harder to access what we’re looking for. Topic detection is one of several techniques that can help you search, organize and understand the information. It can help you discover hidden topical patterns in the text as well as links between words. For example: the words ‘soccer’ and ‘goal’ will appear in files about football. The words ‘paint’ and ‘style’ will appear in files about art. However, the words ‘the’ and ‘is’ will appear in both documents.
The ‘topics’ produced by topic modelling techniques will cluster similar words, and allow you to search for groups of words within documents.
Would you like to receive more information about these services? We’re happy to discuss the possibilities for your organization, just send us an e-mail.