Novel Software Assistant for Automating Knowledge Capture

点击翻译按钮获取中文摘要

Knowledge Base Systems (KBS) are vital for knowledge exchange & learning, proven toreduce costs, alleviate workforce shortfalls & stimulate innovation globally. The keylimitation of KBS is Knowledge Acquisition (KA) - extracting expert knowledge in acomputable format. Accuracy of KA is highly dependent on interviewer skill & expert abilityto articulate knowledge, & projects are likely to fail at this stage causing a “KnowledgeAcquisition Bottleneck”.Empiricom has developed, patented & commercialised a novel KA methodology, SOLARAcquire (SA): free from interviewer dependency, it assists articulation via a structuredinterview process using Boolean (closed) questions to produce a computable output thatexhaustively documents an individual’s decision-making & encodes their entire expertise onany subject. Benefits of SA are recognised by global orgs, however barriers prevent wider mktuptake: lacks scalability, labour intensive requiring ~20 hrs of face-to-face interviews withexperts at a high cost (£1k/day).Based on existing customer demand & extensive mkt research, Empiricom have identified aclear business opportunity in the development of a Natural Language Processing KA softwareassistant (NLP KAA), to enable Empiricom’s methodology to be automated thus removingbarriers preventing wider exploitation in B2B apps where a significant mkt opportunity exists(global mkt worth $6.78bn in 2014).However, critical functionality of the KA for B2B apps requires further exploration, toinvestigate the feasibility of a NLP KAA with:- Full automation of the SA method such that, via an interaction with a machine, an expert in acomplex subject area could successfully have “n-dimensional” knowledge captured from them- Natural language communication incl. correct translation of pronouns & grammaticallycorrectsentential reformulations.The project will result in PoC algorithms for the NLP KAA. Further prototyping will followwith expected mkt intro 2017