Search (1 results, page 1 of 1)

0.01782069 = product of:
  0.03564138 = sum of:
    0.03564138 = product of:
      0.07128276 = sum of:
        0.07128276 = weight(_text_:mining in 2733) [ClassicSimilarity], result of:
          0.07128276 = score(doc=2733,freq=2.0), product of:
            0.28585905 = queryWeight, product of:
              5.642448 = idf(docFreq=425, maxDocs=44218)
              0.05066224 = queryNorm
            0.24936332 = fieldWeight in 2733, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              5.642448 = idf(docFreq=425, maxDocs=44218)
              0.03125 = fieldNorm(doc=2733)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

This paper argues that the Semantic Web needs to incorporate both formal and associative structures (and possibly a multitude of other structures and strategies) to be successful. The arguments for this claim are based on an observation of successes and failures in the areas of artificial intelligence (AI) and natural language processing (NLP). 1. Introduction The WWW provides numerous challenges for information and knowledge processing activities. Information may be available but not accessible or retrievable because of language barriers or insufficient search strategies. Data mining techniques may discover implicit information in explicit data but these techniques do not necessarily guarantee that the discovered information is relevant, significant and trustworthy. During the last several decades hundreds or thousands of computer and information scientists have developed probably thousands of natural language processing and artificial intelligence techniques that were aimed at solving problems related to intelligent information processing only to encounter more and more new obstacles along the way. The latest solution, the Semantic Web, appears as an open declaration of defeat: since natural language processing and AI techniques did not provide sufficient results, it is now proposed to put the burden an the shoulder of the authors of webpages who are expected to populate their pages with metadata and additional markup. Metadata is essentially a new form of controlled vocabulary; markup - at least in the form of XML, XSL, etc - is essentially a programming language. Existing studies of the use of controlled vocabularies and indexing practices in information science and studies of teaching programming languages to "everybody" (Python, 2002) have shown that both are difficult and full of unsolved problems. This can further dampen the expectations of the success of the Semantic Web. In contrast to machines and despite numerous inter-cultural conflicts around the world, humans do communicate surprisingly successfully even across national, linguistic and cultural boundaries. The question then arises: why are humans successful at information processing tasks such as information integration, translation and communication, which computers find so difficult? One obvious answer is that human cognition is embodied and grounded in our shared experiences of living in the same world. AI researchers have theoretically explored the idea of symbol grounding in the early 1990's but so far, connectionist artificial agents with perceptual interfaces have not been integrated with a large-scale capability of symbolic representations.