I tried a couple of searches with Google & wolframalpha for population data (mobile phone penetraion in China), and wolframalpha won it hands-down.
Using goolge, I have to
- visit multiple sites before,
- read through some junk,
- kill some annoying pop-up ads,
- scroll down to get to the information, and
- finally interpreting what the author is trying to say
I tried a couple of searches that are tough to fit in a structured format (e.g: paracetamol dosage, flu incidence in US), and Google is much better here (wolframalpha turns up blank page for these queries).
On the negative side:
Wolfram still has to add lot more information to their database, and the natural language query is not very user-friendly. Lot of my queries turned blank pages on wolframalpha, and I found it difficult to create queries for simple questions.
Overall, wolframalpha introduces a great new way of looking at search algorithms - it is not sufficient to present all the webpages that match a particular string, but what is more useful is intrepreting these pages, collecting facts, analyzing them, building a database, and presenting them in a useable format.
I really hope Google works on something similar, and integrates it with their current search. I feel Google can do a much better job at this given its strong financial resources, and vast data that they analyze.
No comments:
Post a Comment