Fuzzy-match searching

By definition, any molecule compared to itself has a similarity value of 100. In this sense, a similarity search can be viewed as a fuzzy exact match search. 

Key concerns about use of name as the primary means of identification of an individual is that it can have many variations that are difficult to be computationally analyzed so as to declare that two or more records belong to the same individual. Fuzzy logic can be used for searching to increase the matching threshold. 

The main challenge here is to search the Phenotype Summary field for results having to do with locomotion. This is a fuzzy sort of problem, best solved using Perl s pattern-matching facility. After scanning throngh the results for a while, we decide to search for any of the following keywords nncoordinated, paralysis, paralyzed, coils, coder, movement, kinky, and jerky. 

While we portray the structure of matter in an hierarchical scheme, we also recognise that searching the structures is done at various levels of precision exact match, non-exact match with false positives allowed, non-exact match with false negatives allowed, fuzzy searches, heuristic searches, and similarity searches. Similarity searches themselves can be done at many adjustable levels, and similarity itself is multidimensional. 

Our molecular property similarity searching involves the fuzzy numeric matching of a query and file substance molecular properties. Figure 42 illustrates a target molecule. 

Figure 48 illustrates a fuzzy structure query (symbolic) and some retrievals from the experimental BRS file via a prototype system. The circle and square labels are used as points of reference to help indicate the user specified search requirements. The structure fragment in both the circles and squares must be present (as opposed to may be and must not be present), but only at a generic match level. The match level for the square is a generic element the CAS generic group, for the circles. The substitution on both the square and circle fragments is must not be substituted (as opposed to may be and must be substituted). It should be noted that the portion of the query structure between the circles has the default values of may be present, may be substituted, and any match level, i.e., the maximum fuzziness . 

The fuzzy name match is a browse function in which the system locates and retrieves chemical substance names that directly match or are similar to an input name. The match is accomplished by an initial search using generic name search keys to obtain a set of candidate names. A similarity metric based on trigram character strings is then calculated between input name and each candidate name and if a pre-set metric threshold is exceeded the candidate names/substances are retrieved in rank order relative to the input name. Two generic keys are used. One key rectifies differences in name format (inverted vs. uninverted), locants, stereo, alphabetical order of substituents, and some spelling differences. The other key has the same capabilities and, in addition, applies vocabulary control via special dictionaries. For example, tosyT is mapped to the more systematic methylphenylsulfonyl . 

---