The big problem with the Internet now is how to get specific, relevant information without drowning in a sea of data. Most present day search engines merely offer you with a list of web-pages based on key-words matches or page links. As a result, each search turns up thousands, if not millions of pages, many of which are totally irrelevant to your need. This wastes time, obscures data and makes it difficult to enhance knowledge quickly.
Some new engines which have started to address this problem (Accoona, Clusty, Teoma, Eurekster, etc.), but the results are still not satisfactory.
My idea is to develop a 'Knowledge Engine', which would be an enhanced or evolved, intelligent search engine with the following features:
Meta-search - There is no need to re-invent the wheel. The knowledge engine would not have to start from scratch, but will use the power of existing search engines to generate the initial list. Some search sites, like Vivisima, Ask Jeeves and Momma ('The Mother of All Search Engines') are already doing this. Probably we could start by using Vivisima.
Clustering - Instead of presenting an unstructured list of millions of websites, I would like to have the results neatly organized in groups of similar pages in folders/clusters. I've seen some search sites do this (like Vivisima, Clusty, Northern Lights).
Configurable, Modifiable, Filterable, Search Within Previous Search Results - With present search engines, you sometimes you turn up a huge list of results. You want to get rid of, say commercial sites, or a particular phrase, or a particular combination of words. My Knowledge Engine would have a feature to allow search within previous results, by adding or subtracting various criteria, phrases, word combinations, categories of sites, by filtering out sites with certain phrases or characteristics, and so on.
Intelligent, Contextual - The Knowledge Engine will turn up results even with different words with similar meaning or concepts. For example a search on the word 'car' would also turn up results for 'automobile', 'convertible', 'vehicle', and other similar words. The Knowledge Engine would also turn up results based on the context, i.e. look for pages in which 'car' is the main topic, and not just a word in passing.
Human referral, directories, mark-up, tagging - Since there are still limits to computer intelligence, the Knowledge Engine would be guided and improved by humans. The developers, as well as the users, would be able mark up or down pages, links or categories as useful, authoritative, commercial, advertising, irrelevant by humans (something like this has been done by Eurekster. Also, sites like Furl, StumbleUpon, De.li.ci.ous, etc. have some good ideas). There could be directories like Yahoo, but created by theusers, to reflect their preferences and needs. The 'Wiki' movement, or ability to edit web-pages, is one way this could progress. Presently web-pages are tagged by their creator, which may or may not help particular users. I would like to have users of the Knowledge Engine be able to attach tags to such pages (not necessarily edit the original web page, of course, but on a separate link), so that more useful pages could be more easily retrieved, and junk or spamming pages are not brought up.
Search of Media - although Google has already taken some steps, my Knowledge Engine would be able to search among different media, like image, videos, music, etc. The one addition I would like to made is the ability of users to mark-up to tag the media, so it becomes easier to find the next time around. Suppose you find a particular image of a beautiful sunset over the Pacific, but the original is tagged with only a place name, say 'Bali'. You could tag it with the words 'sunset', 'evening', 'waterfront', 'ocean', and so on, so you and others could now find it more easily.
Merging, splicing and overlaying information - Sometimes you would like to overlay or combine information from different sources or databases, like names and addresses in one database with occupations or interests in another. The Knowledge Engine would be able to find key words or links between the two data sets and seamlessly link up the information for you. You could also overlay maps with telephone numbers, names and addresses.
Fine tuning, Tweaking - Most people don’t know how their search is performed, and if they do, they cannot change it. The Knowledge Engine would allow users to refine and tweak the search parameters, add or eliminate categories, and otherwise fine tune how it looks for information. You could feed in your profile and interests, and allow the Knowlege Engine to monitor your browsing habits and search history so that it could refine and target its search accordingly to your needs and preferences, instead of bombarding you with useless and irrelevant trivia.
Seamless Desktop, Net and email search - The Knowledge Engine would search both the Internet as well as your hard-drive and email directory.
I'm sure this will be a very useful development, and would be highly profitable as well. Thats why I'm looking for persons who would be willing to join me in this effort, either as programmers or venture capitalists.
Bravenet.com