Infolust is a new context search engine. We deliver inspiration and bigger pictures. It's for all those moments when you sit in front of a website and think "Now that's interesting" and would like to know more. Now this "More" is just a single click away.

The technology behind it is based on System One, an enterprise collaboration and search platform. Over the next weeks we want to tweak and optimize it for larger scale consumer usage.

Bookmarklet

In the lower right corner you see a bookmarklet. Just drag and drop the link in your bookmarks or right-click and save it as bookmark. Now whenever you are on a website you can either just select the bookmarklet and get the context of the whole page, or you just select parts of the page and use the bookmarklet then to restrict the search to that area.

Acknowledgments

Huge kudos to the team, especially rist, Sascha, Martin and Norbert, as well as earl for the initial wikipedia parsing / transformation groundwork. Also to the two main brackets around our secret sauce: Helma, our web-app framework of choice, and the great solid pieces coming from the Lucene Project. And last but surely not least Wikipedia and everbody involved in making it the amazing resource it is.

Known Issues

While the similarity itself has a very high accuracy, search results of this alpha release are strongly varying. The main objective for improvement during this current release is our ability to determine what is actual content on a page and whats just padding / navigation / footers etc. that impedes good results. Furthermore Infolust currently doesn't work so well when you try to get the context of pages that have many different or contrary concepts in them, e.g. news outlet homepages, tables of contents or full blogs instead of single entries. Sometimes providers prevent pages from being fetched at all or deliver alternate pages than the one you saw (e.g. spambot warnings, logins, cookies must be enabled or specific browsers / plugins must be used), in most cases this will trigger a context search for the warning / notification. Pages that aren't publicly available or behind a paywall can of course only be accessed by selecting the content manually and using the bookmarklet. Currently Infolust supports only English and German texts.

About & Next

Infolust is currently provided as limited alpha within our labs (where you can also search images by sketch or download Wikipedia in RDF). Based on the optimizations of the coming weeks it will most likely move on to a more reliable beta and if thats well received we will work on implementing the gazillion of your and our further ideas in a separate spin-off. That said we'd be more than happy to hear from you via team@infolust.com or the makeshift section on our corporate blog.