Personalised Search

Kshitij Chandan writes:

1. ID Mechanism:
Yahoo made it infamous! One ID, many services. Convergence play, carry your settings wherever you go. The next search engine’s personalized configurations for a user should be transportable and ID mechanism is the only way I currently feel can achieve this. So like Amazon’s a9 search. Log on and keep your settings active. Yahoo and MSN are just a step away from this, Google has to start IDs soon. Well seems like it already has maybe.

2. Search history and Drill down searches:
This may sound already ‘in the fray” to many. Search within results is already done by many right? Well that is just the basic of what I have in mind. Data has to be classified into many verticals by the search engine itself and along with the initial search results, give the users options of how they can be filtered out or grouped. Like a search on an artist might throw up groupings based on Shopping for albums, Searching for Lyrics, Fan sites, Chronological grouping etc. Filters might be site based, country based or even Page Rank based where Page Ranks are influenced by the user too e.g. I treat CNET’s articles as highly important/informative etc. After the user gets into a group, provide a drill down. Grouping on chronology might be further drilled downed and wrapped up as Decade-Year-Month-Week-Day. Groupings on Shopping might be further classified on particular vendors, geographical locations etc. Also search history to go back and forth and ofcourse the famous – “I saw that link yesterday, now I dont remember the search string”. Bookmarking is also useful.

3. RSS feeds and AI:
Blog generate loads of content. While users may often subscribe to many, why not provide a way for them to use your search engine to read them. The latest from their subscriptions, filtering for them might be good services. Also classifying the huge data (not how the blogger classifies it) requires great AI. Topix is a start. Further down one might keep track of the likes and dislikes of a user (perhaps by pages rated by them or content filtered out by them) might be useful to classify/auto filter data relevent for the user. AI is the basis of future classification and the less the user has to do to make it work, the better the AI algorithm used.

4. Desktop not the only market.
Almost all users have WAP browsers in their cells and with Java getting embedded, users use them more n more for entertainment stuff. If the WAP world is targetted by the biggies, customers will search while on the go, and those searches would normally very high localized, personalized, quickly delivered content. Also unlike the desktop, people may be even willing to pay for mobile searches, something they are already doing for games/ringtones/WAP access.

Published by

Rajesh Jain

An Entrepreneur based in Mumbai, India.