Linked by Thom Holwerda on Wed 15th May 2013 23:03 UTC
Google "During its Keynote today, Google announced new features coming to its flagship search function - you know, that thing we all started using Google for. VP Amit Singhal spent some time discussing what Google's search functionality will eventually morph into. Google's strategy is summarized by three words: answer, converse, and anticipate. Singhal explained that many of the pieces of these upcoming changes can already be seen in products that Google has recently introduced - namely, Google Knowledge Graph and Google Now, with perhaps a splash of Google Glass, too." I hold on to my hat every time Google changes Search. It's such a vital product in my daily life.
Permalink for comment 561759
To read all comments associated with this story, please click here.
RE: I don't want
by Alfman on Thu 16th May 2013 04:07 UTC in reply to "I don't want"
Member since:


"If I could afford it I would run my own database-like search-engine and open it to the public."

I'd be very interested in working on one too. I'm pretty tired of all the emphasis google has placed on textual search. Yeah I get that's it's simple and what most people want, but it's annoying when I know what I want and I'm forced to water down my intended search criteria into a rudimentary text search.

Sometimes you want a better criteria based search engine. If I'm searching for news, I could filter down to 50 miles from my home. If I'm searching for technical support on an issue, I conduct a word-wide search and filter down to specific bugtracker criteria such as OS/hardware/versions/etc. If I'm searching for a bicycle, I can specify criteria on fields that would show up on the manufactures product page, and include all vendors who sell the product sorted by prices. In other words, something similar to SQL, but for web searches. And before anyone complains that it'd be too difficult for regular users to use - I don't really care, it would be targeted at power users who's motivation to learn it would be having access a wealth of highly structured information.

This has been in the back of my head for a while. Though it would be infeasible for a person of my resources to build a web spider with sufficiently deep indexes to be useful. It still could be possible to build a "search agent" which would conduct the searches across thousands of websites in response to user queries. So instead of conducting searches ahead of time like most search engines do, the searches could be conducted in real time (by downloading manufacturer specs/news/product pages and matching criteria on the fly). What's more, these agents could be persistent, and continue to research results over time until I desubscribe from them.

Anyone who's searched for products on ebay knows how infuriating that experience can be, having a search agent wouldn't be all that dissimilar from having a secretary doing the grunt-work and compiling results in a normalized form for easy review. It'd require elements of AI, which is why the project seems so appealing to me.

I recently invested in some colocated server capacity for clients (anyone need a US VPS?), so I'm thinking of revisiting this project and running it from my new servers.

Reply Parent Score: 3