Harnessing Wisdom of Crowds in Semantic Models of Networks∗

Link mining, social network and semantic network construction are data fitting problems for which relevant information content is encoded in language and names. Norbert Wiener [7] noted that those features that distinguish human communication are “...(1) the delicacy and complexity of the code used, and (2) the high degree of arbitrariness of this code.” The semantic aspect of language is concerned with meaning, making the interpretation of language and names inherently ambiguous. Language and naming are powerful means for describing social and political patterns and relationships. The routine use of data mining and link analysis algorithms to discover significant relationships or patterns does not usually succeed. A large portion of these failures is due, not to the algorithms, or the data, but to the missing human element of interpretation. Automated data mining algorithms cannot replicate difficult and critical thinking. While ontologies provide a natural encoding mechanism, they are static, non-adaptive linguistic filters. Existing ontologies do not support analytical communities of practice in the process of learning and discovering patterns in linguistic data. In a traditional application, an ontology is a product of consensus of a small set of SMEs and knowledge engineers, whose task is to reduce and regularize the vast concept space of the subject matter into a finite set of well-defined concepts and relationships. The reduction process is laborand time-intensive, and is considered final there is no well-understood feedback and error correction mechanism. As a result, the reduction process creates ontologies that do not effectively reflect the data present in the world and thus are of a limited use. A proposed solution addresses these problems by subtly and consistently integrating humanand machinedriven processes of ontological development, refinement and application through the use of public goods theory and harnessing the ”wisdom of crowds”.