In the beginning … there is customer discovery.
As we developed BrainDocs for eDiscovery and put it in front of customers, they found uses for it in market analysis, medical research, medical coding, knowledge management, recommendation engines and strategic planning. Even customers originally looking at the Nathan API for development, asked if they could buy the demo tools.
We learned that users need text analytics tools that allow them to build content analysis into your BI dashboards and reports. So we’re adding the BrainBrowser and BrainView applications to BrainDocs to create the Analysts Toolbox, putting it in the cloud and building integration with big data sources, databases and your favorite BI tools like Tableau and Qlik.
The first release of the Analyst Toolbox will be a trial version of BrainBrowser coming to the cloud in December 2013.
Over the past three years, ai-one engineers have developed reference designs for demonstration, proof of concept and testing. These applications are used to demonstrate the value proposition of ai-one’s core (Nathan) and show developers how the core can be used with other technologies such as NLP to solve language problems. The three applications used by the ai-one sales team are BrainView, BrainBrowser and BrainDocs. We routinely provide the source documentation for these applications (reference designs) to our developer licensees.
As a result of demand from ai-one customers and prospects, we are merging these applications into one browser based user interface and moving them to a cloud architecture with the new Nathan API. This so we can offer them to end users that do not have the time or resources to develop their own.
The three applications included in the ai-one Analyst Toolbox are:
- BrainBrowser – enables users to analyze a document (or any piece of textual content on Box) and “find something like this”
- BrainDocs – uses personal intelligent agents for finding and organizing content on Box by concept “idea”
- BrainView – create big brains from content or social media (think Box documents, user comments and review notes) to explore patterns and sentiment visually
Problem: You are looking for a particular idea (or concept) that could be worded in a variety of ways and possibly buried among thousands of pages of content. Although you may know a few keywords, or tags, to help search for documents; you still ends up reading through dozens (if not hundreds) of irrelevant documents trying locate that needle in the haystack.
Solution: ai-one Analyst Toolbox
Absolutely state of the art, all of the tools in the ai-one Analyst Toolbox are based on the Nathan API, ai-one’s core technology for language applications. Nathan is a new form of biologically inspired neural computing that processes information in the same way as the brain. Unlike other approaches, our API enables machines to learn with or without human supervision. Our technology automatically generates a lightweight ontology (LWO) that detects all relationships among data elements. Learning occurs at the time data is ingested — so it is very fast compared to other approaches. The features of the Nathan API are as follows:
- Self-optimized information processing
- Multiple higher-order concept formation (multiple high order co-occurrence)
- Autonomic learning via multiple context recognition (unsupervised machine learning)
- Language independent (works at byte level)
The Analyst Toolbox leverages this core technology using both static and dynamic fingerprint techniques to deliver a set of tools for the analyst working with free text or unstructured data to classify, organize, filter, search and explore in ways not possible with keyword search, natural language processing (NLP), Latent Semantic Indexing (LSI) or other statistical/mathematical tools. Furthermore, the extraction of concepts expressed in documents into a fingerprint graph enables experts with programs such as SPSS, R and Tableau to include unstructured data in their analytics and visualizations, not possible before now.
The “intelligent agents” that are deployed in the Analyst Toolbox are trained very simply by the user (“find me something like this”) and the list of relevant results are returned in ranked order for the user to review and mark as relevant or not relevant. This process saves hours (=$$$) of review time – and can be a tremendous value proposition to analysts needing powerful text analytics tools that don’t require experts or special training.