Ah, the headline caught your attention :-) Well, don't worry. LSI is not a new fancy designer-drug and although the G company has a history of flying high, I doubt they are on anything stronger than Coke Zero.
But yesterday I just came across this excellent post by fellow blogspot blogger, John Colascione.
In the post he brings some interesting examples on how Google has implemented LSI (Latent Semantic Indexing). Back in 2005 I had the great pleasure of working with Moses Martiny and Kenneth Vester at Mondosoft while they were writing their Master Thesis on one of my favourite topics of all time, Document Clustering. I remember how I through them learned about LSI which is quite an interesting approach to automatic keyword extraction.
With this technique you can get some amazing results of keywords extracted from documents that doesn't even contain the actual words - although it should have!
If I recall correctly the basic approach is something like making a matrix of documents and words containing the entire document collection, and then use an algorithm like SVD to determine the most distinctive words for each document - even without the document containing the words. Funny stuff!
Naturally I couldn't read John's post without trying Google solution myself, and although it's not every term that has good LSI matches, there was some interesting ones. For instance it would seem that the word "~rap" is connected to both "Eminem" and "Lyrics" as well as "Rdf Api for Php" (the last was obviously the most interesting hit in my humble opinion).
Anyway, it's cool that Google is playing around with this technology - just as all the other search giants (and challengers). Now, if only it was incorporated in the search in a better way than the tilde ("~") query line operator.
Tuesday, May 29, 2007
WCF: Duplex is awesome!
One of my first assignments in the new job has been to lookup into a couple of the fun new features in .NET 3.0, like WCF and WWF.
So far, I've found Communication Foundation really interesting, albeit a bit annoying to work with.
It's main force is it's flexibility. Where you would usually have to decide on either building a Web Service, your own custom coded TCP Server or use remoting, you can now just build a standard service and then just put in the configuration file which protocol it should use (like HTTP, TCP, Named Pipes, MSMQ, etc) - or well, at least thats the theory. The downside to this is of course that since there's a lot of stuff thats configurable, you really need to understand all the configuration concepts (like Bindings, MetadataExchange, Endpoints, Security, Contracts, etc) properly and configure it well in order to use it.
Another potential problem could be the performance of this communication since all the communication is handled using SOAP (which means that there's a lot of XML serialization and deserialization going on).
It would also seem that a couple of the problems known from WebServices has been addressed. For instance you now no longer need to put the webservices in IIS in order to use web-services - they'll just open a HTTP port for you and act as their own server. It also looks like error-handling has been improved and it looks like there's now some cross-service exception-handling (although not perfect. An ApplicationException thrown from the server will appear to be a "FaultException" on the client - but perhaps I'm missing something here).
The most awesome feature I've stumbled across until now in WCF is the possibility to make Duplex services, e.g. services that are able to initiate communication with the client.
Sure, you could yourself make each client a service as well as a client and then let them exchange connection information, but now this functionality is build into the communications framework.
Naturally this requires some coding/configuration inconviniences, but once they are done it's easy to implement a state-of-the-art Observer pattern across various machines.
Setup a service, allow multiple clients to call the service to register themself as subscribers to various events, and then let the Service notify them when the event occurs.
Jeff Barnes has put a great article on Codeproject with an example of this.
So far, I've found Communication Foundation really interesting, albeit a bit annoying to work with.
It's main force is it's flexibility. Where you would usually have to decide on either building a Web Service, your own custom coded TCP Server or use remoting, you can now just build a standard service and then just put in the configuration file which protocol it should use (like HTTP, TCP, Named Pipes, MSMQ, etc) - or well, at least thats the theory. The downside to this is of course that since there's a lot of stuff thats configurable, you really need to understand all the configuration concepts (like Bindings, MetadataExchange, Endpoints, Security, Contracts, etc) properly and configure it well in order to use it.
Another potential problem could be the performance of this communication since all the communication is handled using SOAP (which means that there's a lot of XML serialization and deserialization going on).
It would also seem that a couple of the problems known from WebServices has been addressed. For instance you now no longer need to put the webservices in IIS in order to use web-services - they'll just open a HTTP port for you and act as their own server. It also looks like error-handling has been improved and it looks like there's now some cross-service exception-handling (although not perfect. An ApplicationException thrown from the server will appear to be a "FaultException" on the client - but perhaps I'm missing something here).
The most awesome feature I've stumbled across until now in WCF is the possibility to make Duplex services, e.g. services that are able to initiate communication with the client.
Sure, you could yourself make each client a service as well as a client and then let them exchange connection information, but now this functionality is build into the communications framework.
Naturally this requires some coding/configuration inconviniences, but once they are done it's easy to implement a state-of-the-art Observer pattern across various machines.
Setup a service, allow multiple clients to call the service to register themself as subscribers to various events, and then let the Service notify them when the event occurs.
Jeff Barnes has put a great article on Codeproject with an example of this.
Sunday, May 20, 2007
New Job!
After about a month filled with interesting job interviews and exiting offers, I finally decided where I'm going to work!
As of tomorrow I'm working as a Software Architect at Infopaq International.
I find Infopaq to be a really interesting company, mainly because of their ideas and infrastructure. They monitor a lot (!) of media, write news resumees and distribute the relevant news to their customers, along with doing some really fancy media analyzing. In other words, they seem to have tons of data and information to play around with - something thats always fascinated me. I'm looking very much forward into taking the step from Information Retrieval into working for a genuine Information Provider.
On top of that they seem like a fun company with a lot of clever guys and with some great ideas and visions. I can't wait to get started and learn all about this new company in details.
As of tomorrow I'm working as a Software Architect at Infopaq International.
I find Infopaq to be a really interesting company, mainly because of their ideas and infrastructure. They monitor a lot (!) of media, write news resumees and distribute the relevant news to their customers, along with doing some really fancy media analyzing. In other words, they seem to have tons of data and information to play around with - something thats always fascinated me. I'm looking very much forward into taking the step from Information Retrieval into working for a genuine Information Provider.
On top of that they seem like a fun company with a lot of clever guys and with some great ideas and visions. I can't wait to get started and learn all about this new company in details.
Wednesday, May 16, 2007
Playing around with Embedded Objects in IE
Yesterday I felt like playing around a bit with embedded objects in IE - you know, showing Windows Forms in a browser. An old .NET trick that I've used a couple of times.
For some reason it has never quite become the market standard it was supposed to (probably because it's ugly, inefficient and very browser specific) - but it would have been nice with a good alternative to java applets!
I made a quick adaptation of the SudokuSolver from my previus post to see how it would work as an embedded object. This is what I did:
If you are watching this blog in IE, and you have .NET 2.0 installed, and your security settings is just right, there is a chance that you might actually see the Sudoku Solver here:
For some reason it has never quite become the market standard it was supposed to (probably because it's ugly, inefficient and very browser specific) - but it would have been nice with a good alternative to java applets!
I made a quick adaptation of the SudokuSolver from my previus post to see how it would work as an embedded object. This is what I did:
- Made a Windows Forms Library project
- Made a Windows Forms User Control
- Moved the UI from the Sudoku application to the new User Control (as well as the Code Behind)
- Compiled and put on a web-server
- Made a HTML Page that includes it as an embedded object, and put the page on the same webserver (this is important)
<object id="SudokuControl" height="240" width="206"
classid="http://www.thraen.dk/Download/SudokuWinLib.dll#SudokuWinLib.Sudoku">
</object>
If you are watching this blog in IE, and you have .NET 2.0 installed, and your security settings is just right, there is a chance that you might actually see the Sudoku Solver here:
Subscribe to:
Posts (Atom)