We leverage Machine Learning and Deep Learning algorithms to help navigate and pivot a large set of textual data. Built on top of our proprietary Bayesian Neural Network and Generative Model, Signals platform dynamically identifies semantic topic groups based on the context in your input data. This is a complete automated process; however, you can also train the engine with your feedback, more details please "How to Provide Feedback to Signals Engine".
More specifically, this is a three step process:
Step 1 Signals engine starts by performing NLP in over 24 languages. In this step, your input documents will be tokenized into corresponding N-Grams ( N>=2), lemmatized (sort words by grouping inflected or variant forms of the same word), stemmed, removed Junk and stop words, extract Part-of-Speech information, parse our Entity Information, and other our internal process. A large N-Gram-based content network is created based on your input data files.
Step 2 Signals run a Multi-Model approach on top of the N-Gram-based content network. This including using our proprietary text analytics algorithms extended from Bayesian Neural Network and Generative Model, LSTM (Long Short Term Memory), Seq2Seq NLU, and etc. In this step, data input is clustered in semantically meaningful groups. The groups are generated and visualized by statistical significance (e.g., the % attributed to each topic category in the Semantic Category Visualization). Each category is tagged with top representative terms in Buzzwords. Semantic categories aims to tell the data story from the ground up without filtering lenses. Analyzing semantic categories enables decision-makers to take a proactive approach to solve business problems by identifying emerging themes and asking the right questions.
Step 3 Signals automatically processes all spatial (Where), temporal (When), contributor (Who), and other structured data. It will correlate all this information with the N-Gram-based content network for you to pivot and construct analytics questions against your dataset. More on this you can read "How do you process Structure Data?"