Sensemaking

PARC's sensemaking technologies build on deep competencies in natural language processing and information visualization & interaction

Individuals and organizations are regularly overwhelmed with the massive and diverse content—from databases, documents, e-mails, books, websites, newspapers, blogs, articles, reports, and so on—that is necessary to make decisions or take action.

However, understanding this content and making decisions based on it (especially in mission-critical situations) is not just a simple matter of consuming information. To effectively "make sense" of large, heterogeneous, and often unstructured content collections requires:
-	efficient, accurate, and context-based ways of extracting, filtering, and summarizing information;
-	better and more meaningful ways of organizing, visualizing, and interacting with the information;
-	faster, more objective methods for investigating hypotheses, detecting trends or patterns across multiple sources, and otherwise analyzing or interpreting information.

PARC’s "sensemaking" approach and technologies meet these needs, enabling deeper understanding, better decision making, improved communication and collaboration, and greater productivity.

Application Areas

Specific PARC technologies can be combined for a variety of complex sensemaking and decision-making tasks. Sample application areas include:

• Sensemaking for Business Intelligence and Financial Analysis
  • beyond traditional business analytics
  • combined structured and unstructured content analysis
  • automated tagging of content and metadata generation
  • integrated offline + online research sources — for more thorough competitive intelligence and brand management
  • comprehensive customer profiling — for enhanced customer-relationship management (CRM) and similar tools
  • forensic accounting and compliance monitoring — for industry regulations such as Sarbanes-Oxley
     
• Sensemaking for Search Engines and Internet Portals
  • content-based results and user-sensitive navigation
  • semantic, personalized, and contextualized search — for more relevant and deeper content-based search results
  • social search — tapping into collective knowledge and associated online communities
  • informative, as opposed to merely indicative, search result summaries
     
• Sensemaking for Legal Applications
  • accelerated understanding of a large, dynamic corpus of documents
  • assisted evaluation and review — for patents, contracts, technical documents, and other case research materials
  • automated extraction, tagging, categorization, and summarization of relevant information — such as claim data, statutes, rules, clauses, and litigation parties
     
• Sensemaking for Healthcare and Biosciences
  • dissecting complex healthcare situations
  • automated pre-processing and documentation assembly — for health insurance claims and fraud or liability investigations
  • efficient and better research to streamline R&D processes and costs

PARC's Approach

• Draws on deep technical and theoretical expertise in natural language processing, user interfaces, and information visualization and interaction
• Interdisciplinary perspectives span areas such as artificial intelligence, cognitive science, computational linguistics, ethnography, psychology, and statistics
• Enables multiple levels of meaning extraction — from superficial language analysis to deep knowledge representation
• Emphasizes HII (human-information interaction) not just HCI (human-computer interaction) — focuses on how users interact with information not only devices
• Allows users to control and fluidly interact with rich-media displays
• Incorporates PARC security and privacy research ("intelligent redaction") — to develop secure, efficient, semi-automated, and deep content-based ways of protecting privacy
• Builds on decades of influential PARC-pioneered contributions such as the GUI (graphical user interface), FSM (finite-state morphology), LFG (lexical functional grammar), and 3-D visualization and navigation technologies
• Optimizes for user-valued features
• Is informed by recent PARC intelligence-analysis and question-answering tools created under U.S. government research programs

Underlying Competencies

Sensemaking systems usually combine natural language processing/text mining technologies on the back end, with information visualization and interaction technologies on the front end:

• Natural Language Processing for Sensemaking: entity and entity-relationship extraction; content-in-context analysis; semantic text annotation; interest-specific text condensation for informative summarization and notetaking; fact-matching; automated multilingual translation

• Information Visualization & Interaction: large complex information environments; 3-D visualizations and navigation; interactive and interest-specific content exploration; focus + context display techniques; degree-of-interest trees; proximal search; platform-sensitive content delivery (e.g., mobile phone screens); analysis of competing hypotheses

This work has been partially funded by the NIMD (Novel Intelligence from Massive Data) and AQUAINT (Advanced QUestion and Answering for INTelligence) programs in the U.S. Intelligence Community's/ DARPA's Advanced Research and Development Activity center.