Terms file

Standard organiztion term data is included, which may be replaced or extended with site-specific custom data.

Terms file entries do not contain variants. Multi-token terms with variant forms are defined in the organization variant file. For example, the term "REAL ESTATE" has the variant "RL EST". Each value is defined in a terms file and as variants of each other within the organization variant file.

The presence of either terms or variants will cause organization names to be parsed into multi-token phrases for search and comparison purposes. This means phrases like "TRACTOR TRAILER" will not compare favorably to "TRACTER TRAILER" even though the two phrases differ by only one letter, as the second phrase is neither a term nor a variant and will be parsed as individual name tokens.

Term data is provided in standard .ini file format.

[version]

The version section exists only for validation. Any version data supplied is acceptable.

[terms]

Actual term data appears in this section, in the following format:

termtext,weight,culture:culture

termtext

Text to be treated as a term.

Term text can be single-token or multi-token. Terms will most often consist of multi-token text. A multi-token term is sequence of two or more words (text with embedded white space) that together refer to a single concept.

weight

Specifies the relative contribution a term makes to a match score.

Term weighting can have confusing effects, so the recommended weight value is 1.0.

A weight of zero (0) indicates an existing term should be removed.

culture(s)

A list of culture codes associated with these variant entries, separated by colons. The single character 'A' can be used as shorthand for all cultures.