p2pnet view P2P:- Linking and searching are arguably the net’s two most important characteristics. Without them, the net would be merely a disparate system of disconnections.
The first post sparked a Reader’s Write from Beniz.
“You could use Seeks instead, http://www.seeks-project.info to build your own scroogle, and beyond”, he said. “I mean, why should we depend on a closed source solution, when we can build our own, and deploy it at will ?”
This comment inspired the second post in which p2pnet quoted Emmanuel Benazera (right), aka Beniz, on “the main rationale behind the Seeks project”.
We asked Emmanuel for an update. In it, “As crucial as web search is today, it is clearly lagging behind the other services, allowing no interaction, no feedback, no sharing of any kind”, he says.
Here’s his Introducing the Seeks Project, an Open & Collaborative Decentralized Web Search Engine >>>
Conventional web search is a solitary place. Searchers issue a short set of keywords to a remote black box search service and get a list of URLs as result, wrapped up with few additional information. Hundred of millions of searchers issue hundreds of millions of queries every day. While all those queries are grounded into real life events and activities common to many searchers, users remain isolated from each others.
The ‘Seeks Project’ (http://www.seeks-project.info/), builds an Open Decentralized Architecture and a Free Software application with the mission to create a collaborative and social web search engine.
It is likely that collaborative tools around web search will soon grow out of many services and places. The Seeks Project may then be one among many to come, but builds on a very simple and natural idea that we believe can regenerate the current state of web search.
This simple idea is to regroup users whose queries are identical or similar so they can share and collaborate on the results from conventional search engines. Later on in its development Seeks can gradually move away from the conventional engines.
As of today, the conventional search engine paradigm appears in opposition to the social turn taken by many of the most popular and empowering services on the web and on the Internet at large.
As crucial as web search is today, it is clearly lagging behind the other services, allowing no interaction, no feedback, no sharing of any kind.
Sharing begins where groups of users are formed. Instead of giving away their queries and forgetting them, users can store and take control of their search history. Seeks allows to do so on a P2P network. On this basis, Seeks is able to regroup users with similar queries without revealing their keywords (1).
Seeks’ general equation builds on three elements:
- The first element is the grouping of users who perform identical or similar queries. Similar queries automatically form groups of users with common interests. We refer to these as ‘search groups’, that can be viewed as social search sessions evolving over an extended period of time.
- The second element is a collaborative filtering capability given to every search group. With it every search group can rework the results from conventional search engines, rate them, enrich them, delete them, and so on.
- The third element is the ability for anyone to directly publish information to search groups. Imagine a blogger just finished a new entry on his blog. He will then devise one or more queries that he believes describe his work in this entry. Next, he will query Seeks and get results while being hooked with one or more search groups. Instead of browsing the search results, he will push his new blog entry to the group. Since users are armed with filtering tools, they will process this new piece of information. Additionally, note that this mechanism builds an alternative to crawlers for instant content discovery.
When putting these elements together, what we get is Seeks’ service: a collaborative (the groups) & fair (direct publishing to the groups) real-time (direct publishing + filtering) search engine.
Let us review a series of use cases for this engine:
- You type a query in and get a list of URLs as results, as usual, but get passively connected to one or more search groups. Now you have the ability to re-rank results, rate them, annotate them. This may sound like a rich bookmarking system, but while this information is stored on your machine, you can share it passively with other search group fellow users. This creates a basic collaborative filter on top of the groups.
- You type a query in and do not see anything interesting in the results. Instead of rephrasing your query, you can gradually extend the similarity horizon of your search on Seeks. Roughly, this is equivalent to performing queries of the form “word1 * word3″ where * stands for any word. This enlarges the radius of search groups you get connected to and the number of results you get. The more the similarity horizon is expanded, the more search groups you are in connection with, the more results you get, the more results are far from your original query.
- You type a query in and get other related queries from users belonging to the search groups you get connected to. This allows you to reuse the searching experience of others, through the queries they tried before you when they decided to reveal them.
- You type a query in and get Q&A service or even live chat with people from the search groups you get connected to.
- You type a query in and publish content directly to the search groups you get connected to.
- - You type a query in and get an RSS feed out of the search groups you get connected to. This allows you to stay up to date with the information being published on those groups, 24/7.
Strikingly, such engine does not exist yet whereas it is based on a very simple idea. ‘Wikia Search‘, now dead (2), shared some of the project’s objectives (3). There may be a reason why collaborative web search of this kind has not been embraced by today’s leading players.
Consider that conventional engines feed on user personal data in exchange for a service. These data are used to target advertising and generate revenues. So sharing queries and allowing direct publication to search groups would allow advertisers to bypass the retention centers that conventional engines are. Surely, this is in opposition to conventional engines’ business models.
Keeping critical data out of the hands of the users has been the main business model of many services for long. Recently rising open services have demonstrated that users can develop powerful ecosystems, based on the sharing of data. Wikipedia, Open Street Map, Wikileaks and others are showing the way.
Considering web search, Seeks builds on a series of strong principles, and implements technical solutions to meet them. The five main principles are below.
- The take back of web search requires user control of web search algorithms. Seeks implements the automated grouping of users.
- Web search of tomorrow requires transparent ranking. For this and other reasons as well, Seeks is Free Software.
- Web search of tomorrow requires fairness in content deployment. Seeks meets this by allowing direct publication to search group of seekers.
- User and data privacy protection. Seeks builds a decentralized, P2P architecture in which anyone can set up its own web search node to participate in the network, and in which queries are shared while not being revealed (1).
- Web search should not impose a separation between information and advertising. Seeks’ articulation of collaborative filtering and direct publication to search groups forces the adequation between queries & content, being advertising or not.
As of today the Seeks Project implements a Free Software meta-search engine. The decentralized architecture along with search groups and a set of basic collaborative tools are expected to be deployed this autumn. Direct publishing to search groups should follow soon after that. Our long-term roadmap is publicly available. (http://seeks-project.info/wiki/index.php/Roadmap).
The Seeks Project is ran and developed by volunteers. If you wish to do so, there are many ways you can help us.
Financial help is very much welcome as we are in a crucial phase of growth and technical development. If you can help financially, you can use our donation system or contact us directly.
If you are a developer and have some time to share, help is welcome on many topics, from the user interface to machine learning and sysadmin tasks. If you are no developer or do not have the time, you can help by voting for us on ‘Drumbeat‘, a ladder for a set of innovative projects.
Indirectly, this could help financing the Seeks Project.
Another simple way to help is to start using one of our ‘public nodes’
(http://www.seeks-project.info/wiki/index.php/List_of_Web_Seeks_nodes) and to report bugs and/or features you would love to see.
Seeks is an open project, with much room for creative people and novel ideas. With others (4,5) we are trying to build the open web of tomorrow.
(1) Using a mathematical scheme known as Locality Sensitive Hashing
Thanks, Emmanuel. And good luck.
… and identi.ca
Use free p2pnet newsfeeds for your site. It`s really easy!
Net access blocked by government restrictions? Use Psiphon from the Citizen Lab at the University of Toronto. Go here for details.