Alpha Testing issues

A place to discuss and remember items revolving around the alpha testing of the interface, bugs, difficulties, etc:

Formats

• author page -> better formatting (boxes?) move classes of papers higher up, co-authors, affiliations down. Top thing should be link to "all papers" remove "downloaded 0 times"

Metadata

Search Syntax

• find author ellis, j. r. and not author ellis, j
• find a mele, s and not cn l3
• find cc italy
• find cc italy and ps p and jy 2007

Personalization?

Which personal features we would like to offer to alpha testers? Some we have discussed already, like commenting and reviewing, and decided to disable them. Others depend mostly on login (e.g. groups) and so will be likely hidden too unless we decide on the login mechanism question. But some of the activities related to personalization can be allowed to guests too (e.g. baskets). Hence not all personalization features must be coupled with a login mechanism. This is important to keep in mind when deciding about what to offer and what to hide.

Here is a somewhat complete list of all personalization kind of activities in CDS Invenio:

• user commenting
• user reviewing
• user baskets
• user alerts
• user groups
• user messages
• user submissions
• user search history
• number of reads, "people who viewed this page also viewed"
• number of downloads, "people who downloaded this document also downloaded"
• personalization of the search interface - e.g. number of hits per page

Technically, the price to pay for having some personalization features offered to guests without a login mechanism is (i) a need to differentiate between different guest users, generating different UIDs for every guest, leading to unnecessarily large 'session' and 'user' tables, which is not very crawler-friendly. Moreover, (ii) people coming to the site after N days might see their guest baskets gone if their session expired in the meantime, etc.

I guess I would prefer (1) not to differentiate between guests, and hence to hide all personalization facilities for the alpha, if we do not allow user accounts. On the other hand, if we do allow some sort of user accounts, then (2) we can decide which features to hide and which to allow on a module by module basis (e.g. hide reviewing and commenting, but keep baskets).

(It is to be noted though that the former solution would be a "downgrade" for people used to using baskets in CDS, which represents today some 2,949 registered basket users (1,041 coming from CERN) using 3,978 permanent baskets. It might also be a pity for users coming to check out INSPIRE and seeing that nothing "next-generational" is being offered yet over SPIRES, in spite of our poll and further articles on INSPIRE goals. So even if we hide them for a start, we should plan to deploy them rather sooner than later.)

Salvatore's comments on Personalisation

I think our guiding principle for the alpha should be to reassure the users that they are getting the same thing, just better, and not to scare them with things they might now think they do not want (yet). Therefore I would go for (1) and hide all personalisation.

At the same time I completely share Tibor's point that it would be important to show things which are next-generational and therefore I would try to concentrate on the natural extension of the SPIRES functionality first, and deploy later the personalisation. Let's keep in mind that in the poll personalisation was 4/5 or 5/5 important only for 20% of the respondents (Fig. 5) compared to speed (87%, Fig.5), cited with (79%, Fig.6), and recommendation (46%, Fig.6).

As we have already now out of the box killer speed, the possibility of doing the cited with and some proto-reccomandation (number of reads, "people who viewed this page also viewed"), I would concentrate on those now.

As the following phase, I think we should converge on the account technique, so that we can release the first improvement, which is the baskets which I would very humbly suggest we re-name your personal library to reassure the concerned user that there will not be a check-out to pay for his basket...