The way keyword research is done today is very one dimensional for the most part. The end goal is usually to be as expansive as possible, and many times using Thesaurus or WordNet expansions can suffice.

With the proposed methodology, the researchers are discovering analogous relationships between keyword pairs. I would imagine that it could prove useful to generate new ideas on topics to pursue or articles to write.

What are your thoughts on the additional applications?

I definitely find the concept interesting, where filtering for related keywords such as, seasonal or negative match, would be compelling and time saving especially if done in an automated fashion.

Although even at the “faster” speed of 100 keyword pairs for 6 hours still seems like a lengthy time. Especially when I’ve been working with orders of magnitudes of Millions+ of distinct keyword phrases.

Lastly, I’m curious if you’d think the results of the research would have been improved or diminished if they had used Google, Microsoft, or Ask snippets instead of just Yahoo! BOSS.