WDF*IDF: Use as an analysis aid, not as a text template
What WDF*IDF can do, where boundaries lie and how you use the model sensibly for topic coverage.
The strongest effect usually arises when the topic is not only theoretically understood, but directly checked on real URLs, page types and search intentions.
What WDF*IDF can do, where boundaries lie and how you use the model sensibly for topic coverage. This wiki page arranges the topic practically and shows, When priority arises, which mistakes are often and how you derive concrete measures from them.
Example: A comparison calls a technically inappropriate term
A WDF-IDF evaluation recommends a word that occurs on several competing sites, but does not fit your own offer. First, check search intent and composition of the comparison corpus. Statistical frequency does not explain why a term on your side would make sense.
Use such lists as a research aid for possibly missing aspects. Do not adopt target values mechanically and do not complement incorrect performance. The calculation is a textual statistic, not a complete description of Google ranking systems.
Where does WDF*IDF come from?
At least the WDF formula was developed by information and language researcher Donna Harman. IDF is a value (probably) applied by the search engines. Harman wanted to give the keyword density in a document a value that information science can quantify and thus use as a weighting value.
But if from the Keyword density a weighting value of the word results in the document in question and this document appears with a certain frequency in relation to thematically identical or very similar documents (IDF), can be determined from the two values how important the document and its main topic, named by the keyword, is overall. There is a very complex formula for this, which we would like to neglect. It aims to merge WDF and IDF into one weighting formula.
The formula is so complex because, as WDF, it is not the mere Keyword density rather, it specifies a logarithmically compressed value, which should lead to a higher accuracy.
WDF*IDF in the Contex of SEO
It is possible that search engine algorithms evaluate texts according to WDF*IDF. Donna Harman’s research was actually aimed at weighting texts in scientific libraries so that a researcher or student does not have to read countless sources, but rather finds the document that comes closest to his search intention. The WDF value is the well-known keyword density with its known relevance for SEO.
Whether a search engine, however, all texts with such banal content (including mere sales pages, fashion pages, etc.) according to WDF*IDF in the ranking, may be doubted. SEO experts suspect that
- a) the WDF value says something more than just counting keywords and
- b) the relevance of texts in relation to the competition plays a role.
One could translate this into the general understanding in such a way that a well-optimized text with optimal Keyword density, which either covers a niche topic or treats a frequently treated topic much better than the competition, has good chances in ranking. After all, the WDF*IDF formula is supposed to show that the weight of a document increases the less often its topic has been dealt with so far, because it adds new, useful information.
However, it must always be sufficiently optimized with keywords in order to be found. However, if a document is sufficiently optimized, but its topic has already been sufficiently dealt with (which lowers the IDF value), it has less chance of a high ranking.
Critical consideration of WDF*IDF
Experienced optimizers and authors doubt either the SEO relevance of WDF*IDF or the whole construct itself, which seems to them like a bizarre headbirth. Demands to write texts according to WDF*IDF usually reject authors who work for online marketing. The insertion of keywords in a very specific frequency and thus density, on the other hand, is common practice.