Human interaction, collaboration, how we interact with the environment around us to find information.

Ten years ago, libraries had just got computers, and all you could use them for was the digital card catalogue. But they taught users about the paper catalogue system and the Duey Decimal system. DD system organises by subject.

Over the last 15 years there’s been a paradigm shift from having to really work hard to find information from encyclopedias, almanacs, libaries etc. Now we use keyword search. But this isn’t really enough any more. Trying to find technical, accurate information is pretty much impossible. More results from google book search than from traditional search engine. come full circle.

Keywords are used to classify information – taxonomy versus folksonomy. If you’re not working in the domain and you’re not familiar with the language it’s difficult to look up information. Silos. If you don’t know the search terms you can’t access the information.

Cross-over between domains and disparities between the amount of information in one domain versus another makes it really difficult to find info on minority subjects. Google results swamped by info about the larger domain.

Idea: need to move towards cloud computing – information everywhere. Every site has its own search engine. Everything is still done by keyword – we need a different type of search engine with user participation. While you’re searching, get people to input links between keywords to enable cross-domain discovery. At the moment keyword searches require training: need to know how people phrase things.

Need to look at changing our search interactions to be able to associate keywords with other keywords algorithmically rather than through social means. This is synonym search. But we’re talking about adding context to it. Synonym search algorithms already exist. worio.com – search results based on results from other people’s searches.

Searching for things require you to actually know what you’re looking for. Hard to find scholarly info through general search engines. Need to use the right tools for the job!

If you go to a librarian and ask for books on photography, they will ask questions to determine whether you’re interested in taking photos, finding good photos, designing a camera, history of photography etc. Context is very important.

Wikipedia will ask you about the context of ambiguous terms.
Yahoo directory used to be navigable in this way – top down is too difficult to maintain. However could construct from bottom up – this is the semantic web.

Knowing what you’re searching for is key.

Classifying new technologies and finding terms to discuss new technologies is really difficult. Delicious is designed to tag things with terms – can relate terms to each other as a folksonomy. Should be looking for a search engine that takes  relationships between tags in delicious: something that can rank as well as tag.

Points systems with search: if you can stream searches through the ratings of sites made by people whose opinions you respect would that make your searches more relevant? Delicious networks could do this.

Should there be some element of human return?

Users make assumptions about what they want but don’t tell the search engine this stuff. So they ask for polymer scientists expecting the results to tell them the top scientists in the world but get a page of graduate recruitment adverts. They are disappointed.

Can we learn from user’s search behaviour and use this to help design search engines?
Part of the solution should be education on how to search rather than software solutions.

Attempts to solve these problems are periodically made but don’t get massive exposure: duckduckgo.com

A challenge: don’t use google, Yahoo! or MSN / Bing for a week!

No related posts.

Related posts brought to you by Yet Another Related Posts Plugin.