Semi-supervised Question Retrieval with Gated Convolutions
23017Authors
nan
Abstract
Question answering forums are rapidly grow- ing in size with no effective automated abil- ity to refer to and reuse answers already avail- able for previous posted questions. In this paper, we develop a methodology for find- ing semantically related questions. The task is difficult since 1) key pieces of informa- tion are often buried in extraneous details in the question body and 2) available annota- tions on similar questions are scarce and frag- mented. We design a recurrent and convo- lutional model (gated convolution) to effec- tively map questions to their semantic repre- sentations. The models are pre-trained within an encoder-decoder framework (from body to title) on the basis of the entire raw corpus, and fine-tuned discriminatively from limited annotations. Our evaluation demonstrates that our model yields substantial gains over a stan- dard IR baseline and various neural network architectures (including CNNs, LSTMs and GRUs).1 1
Keywords
Full Content
Proceedings of NAACL-HLT 2016, pages 1279–1289, San Diego, California, June 12-17, 2016. c⃝2016 Association for Computational Linguistics Semi-supervised Question Retrieval with Gated Convolutions Tao Lei Hrishikesh Joshi Regina Barzilay Tommi Jaakkola MIT CSAIL {taolei, hjoshi, regina, tommi}@csail.mit.edu Katerina Tymoshenko Alessandro Moschitti Llu´ıs M`arquez University of Trento Qatar Computing Research Institute, HBKU tymoshenko@disi.unitn.it {amoschitti, lmarquez}@qf.org.qa Abstract Question answering forums are rapidly grow- ing in size with no effective automated abil- ity to refer to and reuse answers already avail- able for previous posted questions. In this paper, we develop a methodology for find- ing semantically related questions. The task is difficult since 1) key pieces of informa- tion are often buried in extraneous details in the question body and 2) available annota- tions on similar questions are scarce and frag- mented. We design a recurrent and convo- lutional model (gated convolution) to effec- tively map questions to their semantic repre- sentations. The models are pre-trained within an encoder-decoder framework (from body to title) on the basis of the entire raw corpus, and fine-tuned discriminatively from limited annotations. Our evaluation demonstrates that our model yields substantial gains over a stan- dard IR baseline and various neural network architectures (including CNNs, LSTMs and GRUs).1 1 Introduction Question answering (QA) forums such as Stack Ex- change2 are rapidly expanding and already contain millions of questions. The expanding scope and cov- erage of these forums often leads to many dupli- cate and interrelated questions, resulting in the same questions being answered multiple times. By iden- tifying similar questions, we can potentially reuse 1Our code and data are available at https://github. com/taolei87/rcnn 2http://stackexchange.com/ Title: How can I boot Ubuntu from a USB? Body: I bought a Compaq pc with Windows 8 a few months ago and now I want to install Ubuntu but still keep Windows 8. I tried Webi but when my pc restarts it read ERROR 0x000007b. I know that Windows 8 has a thing about not letting you have Ubuntu but I still want to have both OS without actually losing all my data ... Title: When I want to install Ubuntu on my laptop I’ll have to erase all my data. “Alonge side windows” doesnt appear Body: I want to install Ubuntu from a Usb drive. It says I have to erase all my data but I want to install it along side Windows 8. The “Install alongside windows” option doesn’t appear. What appear is, ... Figure 1: A pair of similar questions. existing answers, reducing response times and un- necessary repeated work. Unfortunately in most fo- rums, the process of identifying and referring to ex- isting similar questions is done manually by forum participants with limited, scattered success. The task of automatically retrieving similar ques- tions to a given user’s question has recently attracted significant attention and has become a testbed for various representation learning approaches (Zhou et al., 2015; dos Santos et al., 2015). However, the task has proven to be quite challenging – for instance, dos Santos et al. …
Content truncated to 500 words
Labeled Text
Question answering forums are rapidly growing in size with no effective automated ability to refer to and reuse answers already available for previous posted questions. In this paper, we develop a methodology for finding semantically related questions. The task is difficult since 1) key pieces of information are often buried in extraneous details in the question body and 2) available annotations on similar questions are scarce and fragmented. We design a recurrent and convolutional model (gated convolution) to effectively map questions to their semantic representations. The models are pre-trained within an encoder-decoder framework (from body to title) on the basis of the entire raw corpus, and fine-tuned discriminatively from limited annotations. Our evaluation demonstrates that our model yields substantial gains over a standard IR baseline and various neural network architectures (including CNNs, LSTMs and GRUs).1 1
Publication Details
- Paper ID:
- 23017
- Year:
- 2016
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan