KCSL

The Winning Strategy!

The following diagram compares, conceptually, how competing technologies would analyze a document, and how UniFind™ Technology would analyze the same document using its unique algorithms.

While some products use more elaborate means than others, and many claim to understand context and concepts, all essentially collect and index documents, and then analyze these documents as a linear list of words. While they may go so far as to recognize the title, more often than not, it just becomes part of the linear list. The number of times query words appear in the document and whether some or all of the words appear, contribute towards determining the document�s relevance. The distance between the query words found in a document is also a factor. Essentially, the more query words that appear, and the closer in proximity they are to each other, the more relevant the document is deemed to be.

Since sentences, not words, are the major form of communication, UniFind collects, and then, linguistically parses, indexes, and analyzes documents not only in terms of words, but also in terms of sentences. UniFind recognizes that sentences do not end at semi-colons, nor do they necessarily end at acronyms such as "J.F.K.". While query words found in the document may be far apart in a linear list, UniFind will recognize if they are part of the same sentence, and hence the same idea, thus making the sentence and document more relevant.

UniFind also performs completely unique topological analyses of documents that, essentially, look at documents all at once, in their entirety. These analyses take into account not only where the words and sentences appear in the document in terms of proximity, but also whether they appear in the title, the introduction, the middle, and/or the conclusion. This is all done using a dynamically created gradual scale that most often gives greater importance to the beginning and end of a document than to the middle.

The relationship between every query word to every other query word is also calculated multi-dimensionally, as opposed to simply calculating the relationships of the query words on a linear scale. Other UniFind analyses include a statistical, dynamic, automatic, gradual weighting of word importances based on their frequency within a document, and within the entire collection of searchable content. In addition, all morphological transformations of query words are recognized and weighted accordingly.

UniFind uses all the above analyses to produce a list of only the most relevant documents in descending order of relevancy. The documents that are clearly non-relevant to the query, even when they include some query words, are excluded from the list.

While UniFind takes a more comprehensive approach with its algorithmic analyses than do others, it also structures the analyses so that they can be processed in a fraction of a second. Therefore, UniFind delivers unsurpassed performance, while never sacrificing the quality of its search results.