Intellisearch combines various NLP techniques, such as dependency graph extraction, word embeddings, and term frequency-inverse document frequency (tf-idf), to create a sophisticated solution. This solution extracts triples from both the unstructured text in the corpus and the questions, which facilitates contextual search by storing both syntactic and semantic information. To present the most relevant information from the reports for a specific question, the solution employs scoring and ranking algorithms. Furthermore, it retrieves information from different types of content within the reports, including paragraphs and tables (even those embedded as images). Additional features of the solution include Rationale for the Result and Unsupervised Learning. It supports incremental load, scalability, and accommodates variations of questions.

‍