Durham Zoo: Powering a Search-&-Innovation Engine with Collective Intelligence

Abstract

Purpose – Durham Zoo (hereinafter – DZ) is a project to design and operate a concept search engine for science and technology. In DZ, a concept includes a solution to a problem in a particular context.
Design – Concept searching is rendered complex by the fuzzy nature of a concept, the many possible implementations of the same concept, and the many more ways that the many implementations can be expressed in natural language. An additional complexity is the diversity of languages and formats, in which the concepts can be disclosed.
Humans understand language, inference, implication and abstraction and, hence, concepts much better than computers, that in turn are much better at storing and processing vast amounts of data.
We are 7 billion on the planet and we have the Internet as the backbone for Collective Intelligence. So, our concept search engine uses humans to store concepts via a shorthand that can be stored, processed and searched by computers: so, humans IN and computers OUT.
The shorthand is classification: metadata in a structure that can define the content of a disclosure. The classification is designed to be powerful in terms of defining and searching concepts, whilst suited to a crowdsourcing effort. It is simple and intuitive to use. Most importantly, it is adapted to restrict ambiguity, which is the poison of classification, without imposing a restrictive centralised management.
In the classification scheme, each entity is shown together in a graphical representation with related entities. The entities are arranged on a sliding scale of similarity. This sliding scale is effectively fuzzy classification.
Findings – The authors of the paper have been developing a first classification scheme for the technology of traffic cones, this in preparation for a trial of a working system. The process has enabled the authors to further explore the practicalities of concept classification. The CmapTools knowledge modelling kit to develop the graphical representations has been used.
Practical implications – Concept searching is seen as having two categories: prior art searching, which is searching for what already exists, and solution searching: a search for a novel solution to an existing problem.
Prior art searching is not as efficient a process, as all encompassing in scope, or as accurate in result, as it could and probably should be. The prior art includes library collections, journals, conference proceedings and everything else that has been written, drawn, spoken or made public in any way. Much technical information is only published in patents. There is a good reason to improve prior art searching: research, industry, and indeed humanity faces the spectre of patent thickets: an impenetrable legal space that effectively hinders innovation rather than promotes it. Improved prior-art searching would help with the gardening and result in fewer and higher-quality patents. Poor-quality patents can reward patenting activity per se, which is not what the system was designed for. Improved prior-art searching could also result in less duplication in research, and/or lead to improved collaboration.
As regards solution search, the authors of the paper believe that much better use could be made of the existing literature to find solutions from non-obvious areas of science and technology. The so-called cross industry innovation could be joined by biomimetics, the inspiration of solutions from nature.
Crowdsourcing the concept shorthand could produce a system ‘by the people, for the people’, to quote Abraham Lincoln out of context. A Citizen Science and Technology initiative that developed a working search engine could generate revenue for academia. Any monies accruing could be invested in research for the common good, such as the development of climate change mitigation technologies, or the discovery of new antibiotics.