nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Metatagger Query

j0ej0e

Hi,

Following extract from Metatagger Studio PDF :

Weak terms can be used in one of two ways in the definition of a term in a recognition vocabulary:
! To force MetaTagger to look at the clues associated with the term in a classification
vocabulary.
! To ensure that a specific term (in a classification or recognition vocabulary) is not
assigned as a metadata tag.
You can designate a term as weak by adding a <weak> element and setting it to "true".

Anyone can provide some insight on how a classification vocabulary can make use of clues?

rgds,
joej0e

Find more posts tagged with

Comments

Migrateduser

For categories that have very common terms as labels or alts, it's usefull to make them weak and require clues. For example, using "Microsoft" would identify any article about any of Microsoft's products rather than an article about the company. If you marked the category as weak and added the clues "Redmond" and "Bill Gates", you'd eliminate many of the false-positives.

Cheers,
Clark

j0ej0e

Hi cbreyman,

Thanks for your reply, i do understand the concept behind it.
But is there anyway to make use of clues in a classifier? Because it seems like theres no way to set it up in MT Studio.

rgds,
joej0e

Migrateduser

Unfortunately not - classifiers do not use clues or other filters. In general, classifiers get the same effect by using a higher threshold provided that there is enough training data (training set > 1000 docs, with > k examples/category) and a large enough k value (> 20).