Text Analysis, as distinguished from data mining, or analysis of structured (pre-formatted and numerical) data, is a process that discovers conceptual and actionable relationships in unstructured texts, mostly prose in a natural language. As information volumes grow to unmanageable extent, and organizations – from intelligence agencies or small or large businesses - look towards unlocking their unstructured information assets, Text Analysis can provide matrices and framework for decision-making.

Text Analysis: Different Paradigms

There are several paradigms within which texts could be analyzed-- from complete automation (mostly based on statistical/mathematical processing – done by traditional text mining vendors) to a totally human intelligence and skill-driven analysis. It is clear that there is a long tradition to analyzing texts. Humanities and literature students and teachers have always analyzed texts, and have gotten hold of underlying relationships that are not evident at the surface level. Computational processing of natural languages has yet to reach that level of sophistication, but there has been a steady progress in this area. One of the skills that such processing might learn from literary analysis is that of finding relationships that lie beneath the surface of the text .

An Aristotelian Example of Word-Relationships

Taking Text Analysisseriously, it might be worth wondering whether something could not be done using ideas of metaphor and metonymy that are available in the public domain. One of the earliest persons to talk systematically about metaphor was Aristotle, whose definition of metaphor, put in modern algebraic terms is: A:B::C:D-if Achilles (A) is known for his Shield (B) and Ares (C) for his Cup (D), because of the invariable association, this invariable relationship between the person and the thing this person carries, one could, Aristotle points out, use cup and mean shield: the Cup of Achilles (D of A= B). This is a matter of relationships, which, surprisingly, are also a matter of metonymy. In as much as the shield is always (contiguous) with Achilles, and the cup is always (contiguous) with Ares, that contiguity (a matter of metonymy in the traditional sense) makes it possible for the Aristotelian metaphor-a relationship of analogy to come into being.

Sales Marketing Intelligence: Is your company looking to buy a Sales or Marketing Intelligence solution? Then its time you analyze the solution from a Text Analysis point of view. A report by K-Praxis on Sales and Marketing Intelligence provides a roadmap for integrating Text Analysis with traditional data mining. The complete report (Sales and Marketing Intelligence: The Need for Integrating Textual Analytics with Traditional Solutions) is available for purchase through InfoSphere AB .

Word Relationships: A Contemporary Example

For example, there is an invariable relationship (of whatever kind) between the Twin Towers Attack (A), and the Al-Qaida group (B), and suppose while reading documents related to this, we find an invariable relationship between say, a Charity Organization (C) and a Mr. John Smith (D), it should be possible to discover, either through the notion of metaphor, or through the notion of metonymy, the relationship between D and A and B. It should be noted that this will not really be an inference, in the sense that we cannot be certain about it, but we will know enough already to want to know more, take action, and verify the credentials of D. In short, such processing will allow the formulation of certain hypotheses, on which action can be taken: it will provide us with actionable intelligence.

Aristotle and Text Analysis

Aristotle is well known as someone interested in classifying things, and experiments carried out at K-Praxis beginning to lead to an apparent similarity between ancient thinking and new computation: the difference between human thought and machine can be reduced just a little if we take ancient thinking and modern computation as our test cases. The ancients could not compute, modern computation cannot do the commonest of human activities: thinking. Be that as it may, it is clear that human beings find relationships in things through relationships in words. Looking for new ways of looking at text processing is leading to development of methods that process texts for relationships, and thus take Text Analysis to a higher level.

All this comes together, finally, in the very notion of analyzing texts in order to arrive at some intelligent information so that one can decide to perform certain actions-- to do this rather than that. Relationship between words is one of the important aspects human thought processes involved in classification, and that is where Aristotle might be great help.

At K-Praxis, this aspect of word-relationship (especially metaphor and metonymy ) and opinions was very useful during our preparation of the Pre-IPO Google Dossier: An Experiment in Opinion Mining : people express their opinions by relating A to B and C to D. It is the task of Text Analysis to generate and establish meta-info about these relationships, which is what K-praxis attempted while preparing the Google Dossier.