Analyzing Result Sets with Mindbreeze InSpire

Mindbreeze InSpire can analyze result sets not only within our technology but with third-party technology as well.

Mindbreeze's prebuilt insight apps come straight out of the box, giving users complete control to select their desired insight app and tailor it to their specific use case. This includes modifying the look and feel of the displayed metadata, functionality, search filters, and results – all customizable on the insight app designer with easy-to-use widgets. Changing visualization is as simple as a drag and drop.

When a user inputs a query, the results returned to them are impacted by relevancy models, the insight apps themselves, and user interface (UI) components. Insight apps and relevancy models are implicitly and explicitly affected by a "Nexus" of context and different user profiles.

Nexus refers to a connection or link of relationships between two or more properties. 

Adaptive Apps and User Interface (UI)

As mentioned, the result sets can be transformed via insight apps based on the role, department, behavior, and user expertise. In addition to the 360-degree view of the information received on any search, our product's ease of building and defining insight apps is exceptionally unique. Anybody who knows the information needs can create display models – absolutely no programming skills are necessary.

Mindbreeze is multi-channel, so not only is it possible to use the result display within the standard entry point, but one can also view it on mobile apps or portal intranets. Also offered are various application add-ins and integrations, including but not limited to:

• Microsoft Outlook
• Salesforce
• Microsoft SharePoint and SharePoint Online

Third-Party Technology

Not only can Mindbreeze InSpire query based on natural language queries. Mindbreeze InSpire can query based on complex structured constraints such as ranges of numerical values, quantities, geological information, or a mix to form even more powerful expressions.

Data aggregators running on a selected set of items or documents give Mindbreeze InSpire the power to build rich analytical results leading to the formation of business intelligence (BI) tools. Mindbreeze InSpire is equipped with export functionality. Export functionality efficiently gathers not only the most relevant results but all results returned from a search. An export view of the data can be adapted via Structured Query Language (SQL). Now, a user can integrate searches as dynamic tables in more traditional BI tools. Our robust semantic content pipeline allows for structure extraction from large objects on the fly.

For example, datasets or excel worksheets can repeatedly be extracted. Now, one can query on individual rows or subjects and not only the document source as a whole – all used to build rich analytics based on the data.

How Can Data be Modeled to Support Analysis?

We recently discussed our robust semantic pipeline. Our so-called semantic pipeline permits the solution to extract structured and unstructured data from documents and queries. This multi-stage pipeline is highly configurable and extensible, allowing for maximum control over how both data types (structured and unstructured) can be extracted and applied to documents and indexed items.

In order, the multiple stages of the pipeline are:

1. NLP Deep Network Models
2. AI Pipeline
3. Taxonomies/Ontologies/Catalogs
4. Rule-Based Entity Recognition

The details of these stages can be seen in a previous blog post, "The Semantic Pipeline – Understanding and Connecting Data."

Each stage can be configured from the low-level representation (binary content) to extract structured information and items related to the content. The simplest forms are content filtering, retaining layout information, language detection, text mining such as text segmentation, part of speech tagging, and named item extraction. There is symmetry between the content extraction and the query processing, allowing the system to detect items in the text and the queries.

Mindbreeze combines different methods and mechanisms such as text classification, entity recognition, predictive models, machine/deep learning, and artificial intelligence for meaning-based computing, feature selection, and reasoning. This multi-stage semantic processing is not only used for document, metadata, and content processing; it also applies entirely to query processing.