Every Tuesday night, AltSearchEngines brings you a discussion between two leading Alternative Search Engines. Tonight’s topic is “Social Search,” and our two participants are Searchles and Wikia.



1. How does social search improve on the conventional search engines currently out there?

In every angle we can possibly think of!

We’re starting by creating an entirely open source platform for search, from the crawling, advanced content analysis, indexing and ranking, and query processing. Not only is it open source, but all data we generate will be available under an open content license, and available via open protocols.

From the other side, from the searchers and the result pages, we’re following the wiki model very closely and allowing anyone to participate immediately to improve the result for the keywords they searched. An open wiki will host “mini” articles for any keyword or set of keywords, and these are displayed at the top of a search result when available. Any search engine can include these and link back to the wiki to empower the users to participate and improve the mini articles.

Social search is meant to leverage human intelligence to create a better search. The sheer volume of content on the web is staggering - web users want to take shortcuts to their relevant content and they want to be alerted to this content as it’s highlighted by their fellow friends and users. Search is about finding desired content and your friends and peers can help.

2. How is “trusted content” integrated into your community search model? What about “spam” issues?

It’s an open wiki and is managed like any wiki, by simply watching for vandalism or inappropriate behavior.

Searchles does not believe that the “wisdom of the crowd” is all it’s cracked up to be. We believe in perspectives, because searching for “jaguar” has many different types of results. By integrating elements of a social network directly into the search, Searchles has created a way for users to build out networks of people whose expertise they trust and this network of trust influences your search. You can also take advantage of someone else’s network of trust by searching through their content, their friends’ content, their groups’ content, or their friends of friends’ content.

Spam is a serious issue with search. With our approach to social search, spammers can be identified and reported, but the system is also setup so that if no one important is connecting to spammers then their content will be less visible.

3. How do your algorithms interact with the added human intelligence? How scaleable is the search model your site employs?

All of the algorithms are open source, so the deepest levels of human intelligence are an open book that anyone can help edit and improve. The platform is also completely distributed, allowing anyone to run different aspects of it independently and inter-operate with everyone else. By being distributed, anyone can collaborate and/or compete to collectively provide quality results. It will scale based on the number of participants in this distributed system.

Searchles allows you to discover content with personalized filtering features that include tags, keywords or a combination of both, as well as the ability to apply these same filters to search all postings, groups, friends, or friends’ friends based on your own criteria. By analyzing the associations and patterns between trusted people, sources, tags and content, Searchles is able to deliver more precise, relevant results while suggesting content and people related to your interests. There is a very bad long tail problem with social search engines that take the approach of editing specific queries. We avoid this by analyzing the tags that people use when they submit content, and use them to help determine relevancy. If a user tags a document with “social search”, “community”, and “open source” then when someone does a search for “open source search community” it can benefit from these tags without hitting on that particular tag exactly.

4. How might your search engine partner with other Alternative Search Engines (or have you?), or do you plan to essentially “go it alone?”

It’s wide open, anyone can participate. By building out an open protocol, nobody even needs permission to work together, they simply need to implement a few standards and they can immediately join the effort.

We’re currently geared towards forming partnerships with original content sites that have valuable content by providing a more engaging and interactive experience for their users through our social search features. However, we have talked with a few potential partners on the ASE list. We listen to the needs of our users as much as possible, and if we think there is a feature or product they will benefit from that requires a partnership, then we are all for that.

5. What is your primary objective; to dominate your “space,” to achieve maximum market share, to market your technology, or some other?

To establish search as an Internet-wide open platform, it should be part of the Web, not an adjacent commercial service that feeds off of it.

Our objective is to be the glue between people and content on the Internet using social search to deliver better collaboration and more precise, relevant results. It isn’t all about finding a piece of content and leaving, the biggest benefit of Searchles is that when you join groups and make friends with peers of similar interests, the content comes to you. Our ultimate goal is to use our own general web indexing technology from Dumbfind to create a social search engine that crawls the web and is influenced by Searchles users. We think this will be the holy grail of search.

Thank you to Searchles and Wikia; the floor is now open for your comments!

test

This entry was posted on Tuesday, August 14th, 2007 at 4:00 pm and is filed under Debates. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

Sphere: Related Content