Traditional keyword-based search methods become ineffective for modern business document analysis tasks. In particular, users would expect the search system to understand synonyms and context, similarly to large public search tools. Especially when they cannot access the data, e.g., in a closed document management system of an enterprise. Search efficiency can be increased through preliminary semantic analysis of texts. This enables a user to apply additional limitations (context filters) that cannot be used with traditional algorithms.
AI-assisted Search is a new text search technology that factors in the content and semantics (meaning) of documents. With the help of Semantic Map technology, the search tool finds the most important contexts for a specific collection of documents and helps a user narrow down the search query easily. This boosts the efficiency and relevance of the search.
Alternatively, the user can search for similarity among documents and find all documents with a related meaning, not based on pure keywords.
When users type words in a search bar, they are looking for data in documents relevant to the topic of their interest.
The phrases that are searched for, however, typically turn out to be more prevalent in publications that are not relevant and personalized to the issue. Therefore, it makes sense that users would include terms that limit the information in the anticipated document group. Such a process of focusing or searching is frequently iterative: users keep including and removing terms until they get the desired results.
AI-assisted Search tries to increase search accuracy by determining the user's purpose and the context in which words are used in the data to get more relevant results. Our custom AI and Machine Learning algorithms analyze large volumes of texts to extract meaning from separate fragments and explore different topics and how they are related to each other and to a user’s context.
Unlike keyword search, AI-assisted Search takes the user's intent into consideration to get at the contextual meaning of terms. AI-assisted Search pushes beyond the boundaries of the organization's understanding of its data base to get at information and concepts that haven't been explicitly written into the query. In two words, keyword search gives users results based on what they said, and AI-assisted Search helps users get the information they wanted. It tries to "think like a human" by analyzing context and synonyms.
For showcasing the technology, we prepared a demo based on English Wikipedia and made several pilots on medical articles, papers on economics, and legal documents. Here are other user types who can also benefit from our search:
We used Wikipedia as a Proof-of-Concept and evaluated the full text corpus to create a navigable map with dots (topics), connections between them, and subject groupings they form based on their semantic proximity.
Beyond the functionality included in the demo, the AI-assisted Search is compatible with other standard search features, like search suggestions or personalized ranking of results.
Our other demo, Briefly, is also using the AI-assisted Search technology to recommend related search keywords and Wikipedia articles.