Hi all, Thank you for giving me the opportunity to work with Xapian :) I am Jiarong Wei, a third year undergraduate student in Zhejiang University, China. In GSoC 2014, I will work on Letor module with Hanxiao Sun. Here are some questions I encountered these days, 1. In letor.cc, we have two parts of functions: the training part and the ranking part. I?ll use SVMRanker as an example. The training part basically uses the libsvm library and training data to train a model, then save the model file. The ranking part will calculate score for each document in searching results (MSet) by using the trained model file. My question is for each of our three rankers: 1) SVMRanker 2) ListMLE 3) ListNet, do we need three different types of training part? (The ranking part for each of those have the same form I think) I?m not sure the parameters for these three different rankers are the same or not (I guess they?re different). In my understanding, the letor.cc basically just pass parameters ranker. It?s the ranker will do training and calculating things actually. So if we can generalize the form for training part, we don?t need functions like prepare_training_data_for_svm, prepare_training_data_for_listwise etc. We just need prepare_training_data instead. (We can benefit from inheritance of ranker in training part just like in ranking part) 2. There is one thing I have to confirm: once we have the training model (like model file of SVMRanker), we won?t train that model again in general. (The behavior of questletor.cc under bin/ confuses me) 3. Since RankList will be removed, according to the meeting last week, its related information will be stored under MSet::Internal. My plan is to create new class under MSet::Internal. That class will have two kinds of feature vectors: normalized one and unnormalized one. Since it?s in MSet::Internal, there is a wrapper class outside it I think. So it also needs to provide corresponding APIs in that wrapper class. Also, the ranker will use MSet instead of RankList. Do you have any suggestions for this part? 4. For FeatureVector, I think it could be discarded since it just stores the information of feature vector of each document, those information will be stored in the new class in MSet::Internal mentioned in 3. 5. For Feature (letor_feature.cc), I think it could be a static class. It mainly focuses on the calculation of different features. For this part, I?m trying to figure out a better method to implement it. In the meeting last week, Olly and Parth suggested using a dispatching function to calculating different kinds of features because different features, like query-related feature and document feature, will use different parameters to calculate. By adopting this method, we should write down every calculating method in the same class, it?s a little hard to extend to use more features. If a user wants to use his own feature, he need to modify our source code instead of adding his own thing and making letor module use it, like implementing his own feature calculation class and call letor module to use it. I just think it?s not that convenient to extend features. In GSoC 2014, I also need to implement a feature selection algorithm so this part I think it?s kind of important, I mean the extensibility of features. 6. For FeatureManager, it will set the context for feature calculation, like set Database, set query and what kinds of features we want. It provides some basic information like term frequency and inverse document frequency etc. Also it will have function update_mset to touch feature information to MSet. 7. For feature selection, I don?t know when to apply this selection. We will provide the features we want to use to FeatureManager. So the feature selection will provide some information like this feature is better so it will have larger weight? Or this algorithm will select subset of features we provide to generate feature vectors? 8. Do we have document about unit test? That?s also what Hanxiao is looking for. 9. For automated tests, my idea is to use some data to test the functionality of letor module. It will also cover different configurations, like using different rankers, to test the functionality. I think I need some help for this part. Can someone give me some advice? Thanks for your help :) Jiarong Wei -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.xapian.org/pipermail/xapian-devel/attachments/20140522/127eb712/attachment-0002.html>
Hi Jiarong, and welcome. For future reference (both for you, and for our other GSoC students), it's best not to batch up communications, but to ask individual questions like these as they come up. I can often respond to a short email straight away: it's taken me a while to find time to sit down and respond to this email. Also, don't forget to update http://trac.xapian.org/wiki/GSoC2014/Learning%20to%20Rank%20Jiarong%20Wei/Journaleach day to say how you're getting on: I'm checking it daily but have seen no updates yet. Remember that we can only help you based on what you tell us, and what code you push. Don't be reluctant to push work-in-progress code to github; it's often easier to discuss problems based around some code you've tried making, even if that code doesn't work or is only a sketch of an idea. Try and be present on IRC when you're working; asking questions as they come up there can be helpful. On 21 May 2014 19:11, Jiarong Wei <vcamx3 at gmail.com> wrote:> Here are some questions I encountered these days, > > > 1. In letor.cc, we have two parts of functions: the training part and > the ranking part. I?ll use SVMRanker as an example. The training part > basically uses the libsvm library and training data to train a model, then > save the model file. The ranking part will calculate score for each > document in searching results (MSet) by using the trained model file. My > question is for each of our three rankers: 1) SVMRanker 2) ListMLE 3) > ListNet, do we need three different types of training part? (The ranking > part for each of those have the same form I think) I?m not sure the > parameters for these three different rankers are the same or not (I guess > they?re different). In my understanding, the letor.cc basically just pass > parameters ranker. It?s the ranker will do training and calculating things > actually. So if we can generalize the form for training part, we don?t need > functions like prepare_training_data_for_svm, > prepare_training_data_for_listwise etc. We just need prepare_training_data > instead. (We can benefit from inheritance of ranker in training part just > like in ranking part) > > In general, I think we need a different training part for each ranker.There may be some similarities in these existing rankers, and inheritance would be a sensible way to avoid duplicating code if so, but we'd like to have a framework which we can extend to a completely different type of ranker in future.> > 1. There is one thing I have to confirm: once we have the training > model (like model file of SVMRanker), we won?t train that model again in > general. (The behavior of questletor.cc under bin/ confuses me) > > I'm not familiar with the behaviour of questletor, but I suppose it'sreasonable to assume that we don't update models after initial creation. It would be nice to be able to do so, but I think many training algorithms aren't updatable. I feel I may be misunderstanding your question here, though. Parth: any comment to add?> > 1. Since RankList will be removed, according to the meeting last week, > its related information will be stored under MSet::Internal. My plan is to > create new class under MSet::Internal. That class will have two kinds of > feature vectors: normalized one and unnormalized one. Since it?s in > MSet::Internal, there is a wrapper class outside it I think. So it also > needs to provide corresponding APIs in that wrapper class. Also, the ranker > will use MSet instead of RankList. Do you have any suggestions for this > part? > > This sounds like a reasonable approach. This sounds like something youcould implement very soon, and that is sufficiently standalone we could try and get it merged to master on its own.> > 1. For FeatureVector, I think it could be discarded since it just > stores the information of feature vector of each document, those > information will be stored in the new class in MSet::Internal mentioned in > 3. > > Sounds right to me.> > 1. For Feature (letor_feature.cc), I think it could be a static class. > It mainly focuses on the calculation of different features. For this part, > I?m trying to figure out a better method to implement it. In the meeting > last week, Olly and Parth suggested using a dispatching function to > calculating different kinds of features because different features, like > query-related feature and document feature, will use different parameters > to calculate. By adopting this method, we should write down every > calculating method in the same class, it?s a little hard to extend to use > more features. If a user wants to use his own feature, he need to modify > our source code instead of adding his own thing and making letor module use > it, like implementing his own feature calculation class and call letor > module to use it. I just think it?s not that convenient to extend features. > In GSoC 2014, I also need to implement a feature selection algorithm so > this part I think it?s kind of important, I mean the extensibility of > features. > > I can't remember the details of this but what you're suggesting sounds onthe right lines. We certainly want to design for easy extensibility.> > 1. For FeatureManager, it will set the context for feature > calculation, like set Database, set query and what kinds of features we > want. It provides some basic information like term frequency and inverse > document frequency etc. Also it will have function update_mset to touch > feature information to MSet. > > Again, sounds plausible.> > 1. For feature selection, I don?t know when to apply this selection. > We will provide the features we want to use to FeatureManager. So the > feature selection will provide some information like this feature is better > so it will have larger weight? Or this algorithm will select subset of > features we provide to generate feature vectors? > > I'd expect the feature selection to select a subset of features: but it'salso very good for it to be able to return information that a human can check over to see if it's making plausible decisions.> > 1. Do we have document about unit test? That?s also what Hanxiao is > looking for. > > We don't have many unit tests; there is xapian-core/tests/internaltest.ccwhich runs some tests that could be considered unit tests. Mostly, our tests are what might be considered integration tests (ie, the apitest). The tests were set up before many of the modern testing conventions became commonplace; it would be interesting to have a wider discussion about how we could make it easier to implement unit tests.> > 1. For automated tests, my idea is to use some data to test the > functionality of letor module. It will also cover different configurations, > like using different rankers, to test the functionality. I think I need > some help for this part. Can someone give me some advice? > > I'm not sure what advice you need; Parth - any ideas here?-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.xapian.org/pipermail/xapian-devel/attachments/20140523/fa98bc56/attachment-0002.html>
Hi Jiarong,> >> 1. In letor.cc, we have two parts of functions: the training part and >> the ranking part. I?ll use SVMRanker as an example. The training part >> basically uses the libsvm library and training data to train a model, then >> save the model file. The ranking part will calculate score for each >> document in searching results (MSet) by using the trained model file. My >> question is for each of our three rankers: 1) SVMRanker 2) ListMLE 3) >> ListNet, do we need three different types of training part? (The ranking >> part for each of those have the same form I think) I?m not sure the >> parameters for these three different rankers are the same or not (I guess >> they?re different). In my understanding, the letor.cc basically just pass >> parameters ranker. It?s the ranker will do training and calculating things >> actually. So if we can generalize the form for training part, we don?t need >> functions like prepare_training_data_for_svm, >> prepare_training_data_for_listwise etc. We just need prepare_training_data >> instead. (We can benefit from inheritance of ranker in training part just >> like in ranking part) >> >> In general, I think we need a different training part for each ranker. > There may be some similarities in these existing rankers, and inheritance > would be a sensible way to avoid duplicating code if so, but we'd like to > have a framework which we can extend to a completely different type of > ranker in future. >Ideally, we decided to have only a single method like prepare_training_file and it would be the responsibility of the Ranker's to interpret the data the way they want for example, pairwise approaches need pairs and so on. The data format we have decided is the standard one and commonly used among Letor community. Example taken from the SVM-rank page ( http://www.cs.cornell.edu/people/tj/svm_light/svm_rank.html). So at this moment I would say, please focus on only one method and remove the others. This should also be communicated to Hanxiao. <line> .=. <target> qid:<qid> <feature>:<value> <feature>:<value> ... <feature>:<value> # <info> <target> .=. <float> <qid> .=. <positive integer> <feature> .=. <positive integer> <value> .=. <float> <info> .=. <string> >> 1. There is one thing I have to confirm: once we have the training >> model (like model file of SVMRanker), we won?t train that model again in >> general. (The behavior of questletor.cc under bin/ confuses me) >> >> I'm not familiar with the behaviour of questletor, but I suppose it's > reasonable to assume that we don't update models after initial creation. > It would be nice to be able to do so, but I think many training algorithms > aren't updatable. I feel I may be misunderstanding your question here, > though. Parth: any comment to add? >questletor is just an example of how the code works. Once the model is trained, you dont need to retrain it unless you really want to. So for the better interpretation you can add a condition in questletor that train only when model does not exist.> >> 1. Since RankList will be removed, according to the meeting last >> week, its related information will be stored under MSet::Internal. My plan >> is to create new class under MSet::Internal. That class will have two kinds >> of feature vectors: normalized one and unnormalized one. Since it?s in >> MSet::Internal, there is a wrapper class outside it I think. So it also >> needs to provide corresponding APIs in that wrapper class. Also, the ranker >> will use MSet instead of RankList. Do you have any suggestions for this >> part? >> >> This sounds like a reasonable approach. This sounds like something you > could implement very soon, and that is sufficiently standalone we could try > and get it merged to master on its own. >I am not sure if you really need to store the normalised feature vector, just a method to normalize should do the job. We should definitely consult Hanxiao when he comes to a point where he sees storing a normalised version will help in some way. Btw, which type of normalisation methods are you talking about? If you refer to QueryLevelNorm ( http://trac.xapian.org/wiki/GSoC2011/LTR/Notes#QueryLevelNorm) then you that is the standard and your featurevector would be like that. Do you mean to further normalise it?> >> 1. For FeatureVector, I think it could be discarded since it just >> stores the information of feature vector of each document, those >> information will be stored in the new class in MSet::Internal mentioned in >> 3. >> >> Sounds right to me. >Okay, sounds fair but also please store the additional information such as score and label as the featurevecor class currently does.> >> 1. For FeatureManager, it will set the context for feature >> calculation, like set Database, set query and what kinds of features we >> want. It provides some basic information like term frequency and inverse >> document frequency etc. Also it will have function update_mset to touch >> feature information to MSet. >> >> Again, sounds plausible. >Btw we decided not to categorise features based on types like document dependant, query dependent etc in the end but we agreed to give user the power to select a subset of features may be in form of a list<Integer> or something.> >> 1. For feature selection, I don?t know when to apply this selection. >> We will provide the features we want to use to FeatureManager. So the >> feature selection will provide some information like this feature is better >> so it will have larger weight? Or this algorithm will select subset of >> features we provide to generate feature vectors? >> >> I'd expect the feature selection to select a subset of features: but it's > also very good for it to be able to return information that a human can > check over to see if it's making plausible decisions. >Both the feature selection algorithms mentioned on the Letor ProjectIdea page are subset selection based. It is one time and happens before the training. These algorithms will give each feature a score that how important each feature is and the user needs to select top N features based on some educated heuristics presented in the corresponding paper and the computational power at disposal.> >> 1. For automated tests, my idea is to use some data to test the >> functionality of letor module. It will also cover different configurations, >> like using different rankers, to test the functionality. I think I need >> some help for this part. Can someone give me some advice? >> >> I'm not sure what advice you need; Parth - any ideas here? >The test concerning to xapina-letor would be mainly focused around the features and the rankers. So what you can do is use a small test collection with a few documents and check if the features calculated are correct or not, the ranking using each ranker is acceptable or not etc. Cheers, Parth.> > _______________________________________________ > Xapian-devel mailing list > Xapian-devel at lists.xapian.org > http://lists.xapian.org/mailman/listinfo/xapian-devel > >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.xapian.org/pipermail/xapian-devel/attachments/20140523/e467406a/attachment-0002.html>