Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
2000-10-02
2003-02-18
Kindred, Alford W. (Department: 2172)
Data processing: database and file management or data structures
Database design
Data structure types
C712S002000
Reexamination Certificate
active
06523026
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to a computer implemented innovation process. More particularly, it relates to the automated retrieval of analogies useful in constructing innovative solutions to problems, and in identifying novel potential applications for existing products. The method of the invention identifies analogous structures, situations and relationships in semantically distant knowledge domains by creating abstract representations of content (vectors) which are characteristic of a given domain of knowledge (source domain) and searching for similar representations in semantically distant (target) domains. The process of this invention does not depend on symbol (i.e., key word) matching.
BACKGROUND
The ability to recognize novel but relevant analogies from other knowledge domains as clues to potential solutions of problems in a given domain has been a valuable skill in industry. It has long been regarded as a central feature of innovative thinking. Likewise, the ability to recognize new and practical applications for existing products has long been regarded as a valuable talent in marketing. People with such skills are expensive to produce, and can generally operate in a limited range of knowledge domains (i.e., those specialist domains in which they were trained, or had experience). Another serious limitation is in the ability to process large volumes of information, in order to identify relevant analogous situations which have not previously been recognized. Raw information must be analyzed in considerable detail, but there are limits to how much information human beings can process effectively. To make matters worse, most of the raw information used by people in technical and marketing fields is in the form of text. Reading of text is an especially slow and tedious means for acquiring new information. This has resulted in the problem of information overload.
Computer implemented processes for information storage, transmission, and retrieval have accelerated the pace of technological change and increased the intensity of competition in business. Such processes have made information of all kinds much more widely available, and greatly increased the speed at which information can be transmitted. However, most present day methods for retrieving electronically stored information rely on matching of symbols (such as key words) and, to this extent, such systems have made the problem of information overload worse.
Different knowledge domains use different symbols, and even those symbols which are common to most knowledge domains (such as the commonly used words in human languages—which account for most of the words in specialist text) can have different meanings in different areas of knowledge. Variations in the meanings of common symbols (such as most words in human languages) from one area of knowledge to another may be radical, or quite subtle. To make matters worse, these relationships are dynamic. As such, it is extremely difficult to pin down, at any particular time, what a given symbol means across a range of situations. Such domain specific variations in meaning, combined with a proliferation of new specialist terms, has forced people to specialize more and more narrowly over time. This increasing specialization has, in turn, accelerated the speciation of new meanings for existing symbols (i.e., words), and new specialist terms.
The trade-off between precision and recall in conventional “key-word” search technology is well known. These types of systems only retrieve records or documents containing exact symbol or word matches, but take no account of context. Unless one searches very narrowly (i.e., for a few domain specific terms, or other specific groups of words in close proximity), one obtains mostly non-relevant material. Because ideas can be expressed in many different ways using different words, a narrow keyword search can miss most of the relevant records. Searching more broadly may require domain specific knowledge on the part of the user (i.e., as to the relevant synonyms for words used in the query, and different ways of expressing related ideas in different domains). Broader searching, however, brings in additional irrelevant material. Searching broadly in several different semantically variant knowledge domains can easily bring in so much irrelevant material that the user is overloaded, and the information retrieved is therefore useless. Another significant disadvantage of systems which retrieve information by symbol matching is that they tend to retrieve only information that is already known. Unless the user is truly an expert in a range of different areas of knowledge, it is almost impossible to use this kind of technology to make connections that are both relevant and novel (i.e. innovative connections). For example, if one searches a large technical database in order to find applications of polyurethanes in telecommunications, one may enter the Boolean expression “polyurethanes AND telecommunications” as a rather broad search strategy. This will retrieve a handful of very well known applications of polyurethanes in the telecommunications field, plus a larger number of irrelevant records (where the two terms co-occur, but in unrelated ways). One might also use index based search techniques such as subject codes, or art groups, but these tend to focus the results even more tightly on the known applications. It is not possible, by these methods, to find records which although highly relevant do not contain both of the terms in the search.
The domains of telecommunications and polyurethanes are semantically distant, although there exists a small percentage of records in the telecommunications domain that do mention polyurethanes (or widely recognized synonyms thereof). It would be a simple matter to edit out these references to known applications (i.e., by employing the Boolean NOT operator, to create a limited set which excludes records in the telecommunications domain that mention polyurethanes). This would result in a very large and, from the viewpoint of a polyurethanes specialist, intractable mass of records dealing with telecommunications. A polyurethanes specialist who wishes to find novel and relevant applications in the telecommunications field would be faced with the choice of:
1) Becoming an expert in telecommunications,
2) Acquiring the services of a telecommunications domain expert (and bringing them up to speed on polyurethanes technology), or
3) Reading (or scanning) thousands of documents on telecommunications, with the hope of finding something relevant.
In practice, the polyurethanes domain expert would be able to take some shortcuts. He could talk to other people in the polyurethanes field who have had more experience in the telecommunications field (i.e., customers of the known applications). This is a variation of Option-2, above. The success of such an approach assumes the existence of contacts who are willing to share their knowledge. The polyurethanes expert could also limit his search of the extensive telecommunications literature by focusing on certain broad categories of applications (i.e. “foams”) which are well known for polyurethanes. This is a variation of Option-3 above, which reduces (but does not eliminate) the chances of finding truly novel and relevant applications. This latter approach may still result in an intractably large body of records which must be read.
The above example is an illustration of the difficulties in finding semantically distant analogies which are both useful and novel, as a means for solving problems and developing new end use applications. In its present form, the process of innovation by analogy is highly dependent on chance associations (i.e., contacts between the right people with relevant expertise; coming upon relevant records “by accident”; seeing a “related” material, or procedure, or apparatus, in a different area of technology, etc.). These chance associations are difficult to control. Hence, innovation is difficult to control. Even the most ardent and well supported eff
Huntsman International LLC
Kindred Alford W.
Peffer Nicole
LandOfFree
Method for retrieving semantically distant analogies does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for retrieving semantically distant analogies, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for retrieving semantically distant analogies will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3122569