Many patent filings and papers from the search engines discuss ways that they might shuffle around search results to try to provide more relevant responses to people’s searches.
Imagine a search engine changing around the results that you see, not based upon the time that a page is published, but rather on some estimate of the importance of a page to you, and how that importance might vary with time and your calendar. It sounds like a tricky proposition, doesn’t it?
Microsoft adds an element of time to search results by introducing a search system that pays more attention to what is happening on your computer and within your company intranet.
This more complex search system could be used to index both search results and information found on a person’s desktop and local network. The search system would pay more attention to the context of searches and add personalization to those searches by building a user profile to distinguish how important different information might be to each searcher.
It might look at calendar entries, emails, documents, and other information in this attempt to define importance, and understanding different events and times related to things like emails and calendars may be essential in providing relevant results to a searcher.
For example, you may be planning on traveling someplace to attend a conference, and information about the conference is contained in your emails and on your calendar. Keywords and search terms from your personal information might be created and used to rerank the search results that are presented to you if they are relevant to those terms.
The patent application is:
Temporal Ranking of Search Results
Inventors: Raman Chandrasekar, Dean A. Slawson, Michael K. Forney
Assigned to Microsoft
US Patent Application 20080027921
Published January 31, 2008
Filed: July 31, 2006
Abstract
An information dissemination system ranks the search results based on a temporal weight assigned to each search result. The temporal weight is an indication of the importance of a user that varies with time. For each search result, the information dissemination system calculates a temporal weight that is based on the temporal proximity of the event that is related to the search result. The temporal weight may be used to re-rank the search results.
An example of how this could work:
You may have two events scheduled in your calendar – a meeting in New York, and a flight to New York for the meeting.
At some point in time before that meeting, the importance of the meeting is high, and search results related to the time of the meeting and the subject involved could be given a stronger weight in search results that you see when you perform searches that might be related to New York, or the topic of the meeting expressed in your calendar and emails.
Information about the flight may also increase in importance, and information related to the flight becomes more important, such as flight delay information.
After you take the actual trip, information about the flight decreases in importance. After attending the meeting, information about the topic and time of the meeting also diminishes in importance.
Conclusion
How useful would a system like this be, that looks at your personal information, and changes your search results around based upon your calendar, your emails, and documents that you might have on your computer?
Does this mean that if you are planning to go to New York during a certain period, that searches for Broadway plays might be shuffled to show you plays that are happening during the time you are in New York City?
Maybe.
A system like this would possibly use a desktop search component, and possibly look in places like Outlook to get information.
People might be searching for the type of information that you suggest already, but this system would try to improve the possibility that they receive relevant and meaningful information.
Automatically serving up information like flight delays, and so on, would be something that a system like this would seem to be able to do.
So let me get this straight, Microsoft wants to deliver search results based on “your time of the month”…
I don’t see how this gives them any competitive advantage in the search engine wars against Google. It seems like they are just using patent filings for search engine technology to grab at straws.
I assume this type of search technology would require some software download like a toolbar for your web browser along with a user agreement that allows Microsoft to look at the contents on your PC.
How else would they do this without upsetting privacy issues and laws? —
Maybe I just don’t get all the brilliant search engine applications for this type of technology. Wouldn’t you expect that a person’s search queries would already reflect the time context when they are doing them? Or is this going to be used for some type of automatic search system?
For instance, if I am planning a trip to NYC in the next week ( or whatever ) wouldn’t I already be searching for events, hotels and travel-related topics for NYC on their search engine? And if Microsoft’s search engine technology is up-to-snuff, it should return the most relevant results based on my searches, not on the information on my company network or PC.
The only useful application that I can see for this type of search engine technology, is if user’s want Microsoft to automatically serve up search results based on what is on their intranets and PC’s.
This type of search system could be similar to an RSS reader, except you wouldn’t subscribe to a particular RSS feed. Rather, MS would automatically serve up RSS type search results based on what is on your network or PC.
Hmmm, the ever expanding search for some kind of advantage in the search engine wars…privacy issues are of course the first concern, as the monitoring of activity on your pc / laptop can lead to all sorts of incriminating analytics relating to your work time activities.
Then one has to question the artificial intelligence behind the algorithms driving this endeavour. Feel that there could be future possibilities here, but will take a back seat and let others be the quinea pigs (same with windows vista…heheheheheh) and debuggers first.
Why is it different when the government does it. But when Microsoft does something similarly creepy it may be overlooked. Guess the ACLU doesn’t know yet.
I am not big on the privacy war debates but someone searching my calender is kind of creepy. It is bad enough google tracks me through the internet…
I must be honest… I really don’t know what to think about this. I am sure some mass media source will eventually break with the story a couple years from now and tell me how I should think about it. Now I am droning on about nothing.
Thanks for the heads up.
I don’t like it myself, but I can see that this technology would enable machines to catch humanlike experience of language use.
In context a word can or a name can mean something completely different. Armstrong for cyclists, for musicians and for astronauts is referring to another person except when the contest of the dialogue makes it extremely clear it isn’t “theirs” they talking about.
Probability for correct disambiguisation of names (of people and places) could be fine to many people.
Interesting result, Gab. Does he have Google’s desktop search installed? The local search implications for something like this are definitely there, too.
Jansegers,
The aim of the patent filing isn’t so much disambiguation as it is context-based search rerankings. But, the personal profile that’s being build might still have problems with the coffee loving computer scientist planning on going on vacation on the island of Java.
You’re welcome Chris. I wonder how much our concept of privacy can change and alter when we see it transformed through public databases that were never meant to be as accessible as they are, as well as information that was never meant to be public accessed easier than it ever should be, plus activities (like browsing and searching the web) that are newer. There is a lot of debate on the subject, but not really in the mainstream media. Places like epic.org are good places to go to learn more.
@ People Finder
My dad did a search on Google.com. The top result in the SERPs came from his desktop; the information is nowhere online. If it’s a privacy concern, and it may well be, the first SE to challenge on it are Google.
And Bill, another use I thought you would pick up on is for ranking local results. Get you a hotel closer to your meeting place ranking higher. FInally a use for proximity in local search.
Does that mean I can’t switch mode at work and watch naughty stuff 🙂
Anyway, I think this could have been very powerful few years back but now most of our work is online anyway. So, are they really tacking about tracking everything we do with our browsers or trying to index all the content within our computer…
I am not ok with either.
I like a bit of hap hazardness in my life …What do you think?
Rajat Garg
I’m not sure what the controversy is here. Search engine companies have been using personal information for years in the form of cookies placed on the local machine that are accessed for search history upon revisit. This history is then used to build a search behavior pattern as well as to inform releveance for results presentation to the customer. Last year, Microsoft deomonstrated the use of an index of local machine files to inform search results presentation with Searching for Michael Jordan… My concern is the indiscriminate use of past behavior to inform unrelated present behavior. The fact that I was looking for a book on gambling as a birthday present for my brother-in-law does not mean that I have developed a new and ongoing interest in gambling. People understand this distinction. Machines and computational math do not.
I like Bill’s example of having a trip to NYC on your calender and the search engine filtering search results on Broadway plays to that time frame. Good illustration of use for this patent unless I am not purchasing the tickets for myself or I have multiple trips in my calender. My wish is that they would all go back to the drawing board and refine the ways for customers to be able to directly impact the search results as their search becomes more informed by what they are seeing. I’m lighting candles for Baysian Predictive algorithms every day. 🙂
What do they have in mind ?
They want to fine tune the search result ? … fine … But first get a _real_ search engine. One that already delivers relevant results by default. Refining results from crap will still be crap. Live search is a subpar engine on all levels:
– Not as many site indexed as the others
– A broken bot that doesn’t understand gzip (what a waste of bandwidth)
– A broken bot that only understand wildcards in robots.txt as DOS 4.0 does at the time.
– Webmaster tools … what tools ?
Lol, the idea is nice but only if you have the option. Useful, perhaps… More a fancy tool to play around with…
@ Rajat Garg, I think that the point behind the process is to make search more convenient for folks, and provide more options. A side effect could be its invasiveness. 🙁
@ marianne, Looking at past searches to try to understand the intent of a searcher, and their interests is difficult. Good points. I think that we are seeing a lot of different ideas being thrown out in patent applications, and in software development.
@ Marc B, There are some issues with Microsoft’s search as it exists now. It’s possible that they may someday improve it. Hopefully we are getting a glimpse of some of the possible steps that they may take to move in that direction by looking at white papers and patent filings from them here. 🙂
@ lening, Thanks. I think there may be some potential here that we may not be seeing. I do like the idea that a search engine should consider the context of a search as well as the content. Maybe it’s a step in the right direction, and maybe it isn’t. One question is how much data we really want to share with search engines to make our searches “better.”
Beyond the privacy considerations pretty innovative idea atleast in theory.
Hi Bull Gang Tactical,
It is interesting to see Microsoft integrating search with desktop applications in new ways. Wondering what’s next. 🙂
Sounds a bit to much like big brother for my liking, that said with global terrorism the way it is I guess all the major search engines work closely with law enforcement agencies, it will all be worded like this patent but the end product is something completely different.
Hi Pete,
It does seem like a big brother element exists within this method. I would hope that any search engine using data obtained from places like the calendar on the computer it is used from would keep that information from being shared by anyone. But who knows for certain?
People might be searching for the type of information that you suggest already, but this system would try to improve the possibility that they receive relevant and meaningful information.How can i use this to my websites ?