22.4.08

Searching in Social Networks vs. Zotero

Every day I discover a feature in Zotero that makes me love it more.

When I came across CiteUlike, I quicky found that the search capabilities of the system were very limited. In fact, there are two exclusive search operations: by keyword and by tag. Searching by keywords is as usual, but very restricted if you are used to Google :-( Unfortunately tags are not better, as you can search for a tag at a time. I miserably miss tag boolean expressions!!!

Correction: I have had 5 minutes more for checking the question mark besides the Search text field, and I have discovered taht the full power of Lucene is behind the scene; that includes tagging searches with metadata (author:Gómez, tag:filter), boolean sintax, and wildcards (simple regexes).

Other social networks are looking for more advanced search paradigms, like Flickr with the new feature "clusters". This feature allows to, given a set of pics filed under a tag, get them orgainzed in groups (clusters). The user selects a cluster and gets its elements again organized into smaller groups. I infer that groups are built according to a image similarity measure based on shared tags. This interaction paradigm was first presented as Scatter-Gather [1], a decade ago.

Regarding tags, I believe that boolean sintax is a must, at least as an "advanced search" feature. Zotero does include the feature, as it has a tag selection box that allows to select two or more tags. Ok, it is a simple "and", but it is something. +1 for Zotero, +1 for Flikr, -1 for CiteUlike +2 for CiteUlike (for being as clever as to use Lucene).

Besides, Zotero advanced search covers boolean keyword expressions on the whole field range (title, author, etc.) with an straightforward interface (click on the advanced search button). Another +1 for Zotero!

[1] Douglass Cutting, David Karger, Jan Pedersen, and John W. Tukey. Scatter/Gather: A Cluster-based Approach to Browsing Large Document Collections, Proceedings of the 15th Annual International ACM/SIGIR Conference, Copenhagen, 1992 (preprint, ACM).