On 2015 Oct 01, Egon Willighagen commented:

Dear Christopher, when you studied the indexing of IKs by Google, did you also look at the effect of SEO tricks? That is, Google suggests that writing a good page title and description will help them correctly index pages. For example, I can imagine you had looked at one of the lists, e.g. for atorvastatin, and determined if the IK was used in the page title and/or description. Would you expect that if chemical databases adopted this SEO technologies for the IK, it would further improve the situation? Or would the effect be minor? Could it help overcome the problem with partly matching IKs, as you noted here about a year ago?

On 2014 Sep 16, Christopher Southan commented:

Unfortunately Google has now become capricious in matching either part of the Key and thus producing false positives from just the second half. It is therfore more effective to try the inner 14-characters of the Key (the connectivity layer) first. The overall utility still stands and the better news is that, even though the full InChI strings are truncated to 32 caracters in a search, they can give useful partial matches.

In reply (Oct 2015) to Egons Q above. As we know, rankings move so its difficult to know what (legitimate) SEO steps are making the difference (exepting traffic per se). Yes, databases could do more, in particular PubChem has a backlog problem (i.e. new entries not indexed). It would also be geat if UniChem and SureChEMBL contrived to get fully crawled. Then we really could check the global 100+ million in a ~0.3 sec pop

On 2014 Sep 16, Christopher Southan commented:

Unfortunately Google has now become capricious in matching either part of the Key and thus producing false positives from just the second half. It is therfore more effective to try the inner 14-characters of the Key (the connectivity layer) first. The overall utility still stands and the better news is that, even though the full InChI strings are truncated to 32 caracters in a search, they can give useful partial matches.

In reply (Oct 2015) to Egons Q above. As we know, rankings move so its difficult to know what (legitimate) SEO steps are making the difference (exepting traffic per se). Yes, databases could do more, in particular PubChem has a backlog problem (i.e. new entries not indexed). It would also be geat if UniChem and SureChEMBL contrived to get fully crawled. Then we really could check the global 100+ million in a ~0.3 sec pop

On 2015 Oct 01, Egon Willighagen commented:

Dear Christopher, when you studied the indexing of IKs by Google, did you also look at the effect of SEO tricks? That is, Google suggests that writing a good page title and description will help them correctly index pages. For example, I can imagine you had looked at one of the lists, e.g. for atorvastatin, and determined if the IK was used in the page title and/or description. Would you expect that if chemical databases adopted this SEO technologies for the IK, it would further improve the situation? Or would the effect be minor? Could it help overcome the problem with partly matching IKs, as you noted here about a year ago?