Using the Thesaurus

FACT-Finder allows you to maintain synonyms, antonyms and other relationships using the Thesaurus.

Synonyms are important in telling FACT-Finder which words are equal to each other, in cases where the FACT-Finder similarity is not able to determine this itself. In other words, when words exhibit a semantic, but not a phonetic similarity (for instance laptop -> notebook).

Antonyms tell FACT-Finder that two words are dissimilar in meaning - in those cases where the FACT-Finder algorithm would otherwise report a similarity or. In other words, when words exhibit a phonetic, but not a semantic similarity (for instance ski -> skin).

Both cases are managed in the same thesaurus list. The difference lies in the reduction in value stipulated between each word pair. Synonyms have a value reduction of 0% (so no depreciation). Antonyms have a value reduction of 100%. The values in between can be used to create word relationships that are not entirely synonymous and yet are also not entirely antonymous (for instance jeans -> trousers). One would reduce the value of this word relationship by a few percent (such as 5%) so that a search for jeans would also show corduroy trousers, but first all pairs of jeans.

FACT-Finder only searches its thesaurus list for exact matches; it has no margin for error. If trousers -> jeans is entered in the thesaurus, and the search is made for "trouser", FACT-Finder would ignore the thesaurus entry.

FACT-Finder processes the thesaurus list recursively. This means that if there is a link between different thesaurus entries, FACT-Finder will detect that relationship.

Example 1

The following thesaurus entries have been made:

Trousers → Trouser

Trouser → Jeans

When searching for "trousers", FACT-Finder would search for trousers, trouser and jeans. This chain can be up to 6 levels deep.

The thesaurus implemented in FACT-Finder is a one-word thesaurus. Relations between phrases are not possible to set up for technical reasons.

Example 2

Trouser → Jeans (OK)

Blue trouser → Jeans (false, "blue trouser" are two words)

Blue-trouser → Jeans (OK, - is recognised as part of a word)

R.E.M → REM (false, the period is a delimiter. This makes R.E.M. into three words.)

Impact

If desired, the Thesaurus makes FACT-Finder treat specific words as similar that would normally not be similar according to the FACT-Finder algorithm. In contrast, words that the FACT-Finder algorithm normally considers to be similar can be marked as being dissimilar.

Changing Settings

You can maintain the word list in the Thesaurus section via the toolbar on the left or via the Diagnostic Search. However, in the diagnostic search you can only work based on search terms. The complete list is not available there.

Recommendation

In general, the value deductions will almost all be either 100% or 0%, since this tool is used to set up synonyms and antonyms. However, it can also be used to record deductions of just a few percentage points in order to display a relationship between terms. Any other values are generally not required.

Based on the construction, synonyms can quickly become synonym chains. It is advisable to structure meanings which are not absolutely similar via slight reductions (e.g. 0,1% or 0,2%). This way a synonym chain cannot disrupt a search result so easily.

Page Contents