Intelligent Systems: HOME

In today's information and content rich world, Knowledge Acquisition activities such as organizing content, developing taxonomies, categories, keywords, and other metadata has never been more important and is the key to customers finding and using your content, especially products and catalogs. Yet this very richness makes developing and maintaining this metadata ever more time consuming and expensive. Developing metadata in a way that matches your customer's vocabulary and the way they think about your content and products is even more challenging.

Intelligent Systems has developed methods and tools to acquire knowledge such as metadata and taxonomies in a scalable way using Crowdsourcing. Some key elements of our approach include:

  • All acquired metadata and knowledge driving the crowdsourcing user interaction itself are stored in a declarative semantic representation stored in a repository using RDF.
    • Parameterized templated questions
    • Answers
    • Users (possibly unique anonymous)
    • Scoring
    • Iterative workflow (see below)
    • Categories (of questions, answers, etc.)
    • Acquired Knowledge (keywords, categories, taxonomies, and other metadata)
  • Knowledge Acquisition is an iterative process whereby answers to initial questions are stored in the repository and organized in a structured way to be used in subsequent questions. This allows refinement of knowledge and acquisition of hierarchical and other knowledge structures such as taxonomies.
  • Statistical techniques to improve quality of knowledge, filter erroneous or malicious answers, and achieve consensus on knowledge collaboratively.
  • Multiple User Interfaces and integrations support most appropriate channel for interacting with users/knowledge workers
    • AWS Mechanical Turk integration
    • Crowd Sorcerer Q/A interface
    • Website integration (see below)
    • Custom
  • Built-in Crowd Sorcerer RDF-based knowledge repository and knowledge browsing tools for inspecting acquired knowledge and its relationship to other knowledge.
  • RDF representation allows leveraging existing knowledge such as existing product taxonomies in knowledge acquisition process and integrating newly acquired knowledge into existing knowledge base. It also allows knowledge transfer and interoperability with 3rd party repositories and systems using RDF, the W3C semantic web standard for knowledge representation, as well as leveraging acquired knowledge in semantic markup to improve search, increase search ranking, etc.
  • Web integration allows interacting directly with your actual customers directly on your website or catalog to answer questions like product or content classification, keywords, etc. This allows you to capture knowledge in the language your customers use and match the way they think about your products or content, which may differ from your product experts or technical team.