| Hi,
i hope you enjoyed your holiday. We're still struggling to find a suitable way to implement our use case in SMILA.
I had a talk with Bjoern that agreed with Jurgen about the possibility to use two different pipelines at query time. Following their suggestions, we would build a first pipeline that is in charge of saving the logos in a given folder, a web application that allows the interaction with the people in a crowd, and a second pipeline that, when triggered by the web app, is responsible of identifying the frames in the video collection that contain one of the stored logos.
If this is the only possible solution, there is still a big question. We want to obtain a result in which the whole process can be seen as a single search pipeline. With the term "search pipeline" we mean the following: if the user looks for "logo1" in the video collection, pressing the button "search" (and waiting for a while) he will obtain the results in the same page in which the search was launched. The problem is: in this hybrid solution that divides the process into two separated pipelines, the first pipeline gets the input from the user, while the second pipeline returns back the results, and they aren't aware of the status of the other pipeline. We suppose that the second pipeline doesn't know how to return the results in the search page. Is there a way of giving back the results in the SAME page in which the search process is launched?
Ciao, Ilio. Il giorno 22/dic/2011, alle ore 18:45, Andreas Weber ha scritto: Hi Ilio, I’m sorry, no experience at all with interrupting and asynch. resuming of a pipeline execution. Best regards, Andreas Von: smila-user-bounces@xxxxxxxxxxx [mailto:smila-user-bounces@xxxxxxxxxxx] Im Auftrag von Ilio Catallo
Gesendet: Donnerstag, 22. Dezember 2011 17:56
An: Smila project user mailing list
Betreff: Re: [smila-user] SMILA vs Human Computing Dear Andreas, we are aware of the sub-optimality of the solution. Unfortunately the solution you propose splits the request/response cycle of the query, thus, forcing us to create a third party application wrt SMILA that orchestrates the sequence of pipeline invocations and hides the asynchronicity to the user. What about exploiting the BPEL engine to introduce asynchronous execution by means of a Receive action as stated in my previous mail? do you have any experience on that? Il giorno 22/dic/2011, alle ore 17:48, Andreas Weber ha scritto:
you can edit the timeout of a pipeline in SMILA\configuration\org.eclipse.smila.processing.bpel\processor.properties (see pipeline.timeout) But as Jürgen wrote before, SMILA is not designed to handle such user interactions within a pipeline. You should really think about the separation of storing/presenting those logos (crowd interaction) from the following video identification e.g. by using different (SMILA) workflows/pipelines. (see Jürgen’s mail) P.S. Be prepared that our list will not be very busy until the beginning of january cause most (all?) people will be on vacation. ;) Von: smila-user-bounces@xxxxxxxxxxx [mailto:smila-user-bounces@xxxxxxxxxxx] Im Auftrag von Ilio Catallo
Gesendet: Donnerstag, 22. Dezember 2011 17:32
An: Smila project user mailing list
Betreff: Re: [smila-user] SMILA vs Human Computing We thought how to circumvent the problem. To evaluate the feasibility of those workarounds we would like to know: - Is it possible to set the pipeline timeout equal to infinity? If so, which is the file that has to be modified?
- Is it possible to use within SMILA a BPEL pipeline such as the one in the attached figure? In this case we would be able to wait for an event sent by the crowdsourcing webapp and let the pipelet P2 modify the content indicated by the users of the crowd. We would like to know the technicalities that stay behind this approach.
_______________________________________________ smila-user mailing list smila-user@xxxxxxxxxxxhttps://dev.eclipse.org/mailman/listinfo/smila-user
|
