Thread-level Analysis over Technical User Forum Data
28647Authors
nan
Abstract
This research focuses on improving in- formation access over troubleshooting- oriented technical user forums via thread- level analysis. We describe a modular task formulation and novel dataset, and go on to describe a series of preliminary classifi- cation experiments over the data. We find that a class composition strategy achieves the best results, surpassing multiclass clas- sification approaches. 1
Keywords
Full Content
Thread-level Analysis over Technical User Forum Data Li Wang, Su Nam Kim and Timothy Baldwin NICTA VRL Department of Computer Science and Software Engineering University of Melbourne VIC 3010 Australia li.wang.d@gmail.com, sunamkim@gmail.com, tb@ldwin.net Abstract This research focuses on improving in- formation access over troubleshooting- oriented technical user forums via thread- level analysis. We describe a modular task formulation and novel dataset, and go on to describe a series of preliminary classifi- cation experiments over the data. We find that a class composition strategy achieves the best results, surpassing multiclass clas- sification approaches. 1 Introduction Online forums and discussion boards are online platforms for people to hold discussions in particu- lar domains. They are widely used in various areas such as customer support, community develop- ment, interactive reporting and online education. Forums provide one of the primary avenues for online users to share information on the Internet. Users post their questions or problems onto on- line forums and get possible solutions from other users. Through this simple mechanism, great vol- umes of data with customised answers to highly specialised domain-specific questions are created on a daily basis. However, it is not an easy job to extract the information latent in the threads. The aim of our research is to help users to more easily access existing information in forums which relate to their questions, by text mining troubleshooting-oriented, computer-related tech- nical user forum data (Baldwin et al., 2010). An example thread from a real-world forum is shown in Figure 1, which is made up of 4 posts with 3 distinct participants. Our proposed strategy is to model the “content structure” of forum threads by analysing requests for information and provision of solutions in the thread data. We devise an ontology of problem sources and solution types with which to analyse HTML Input Code - CNET Coding & scripting Forums User A HTML Input Code Post 1 . . . Please can someone tell me how to create an input box that asks the user to enter their ID, and then allows them to press go. . . . User B Re: html input code Post 2 Part 1: create a form with a text field. . . . User C asp.net c# video Post 3 I’ve prepared for you video.link click . . . User A Thank You! Post 4 I have Microsoft Visual Studio 6, what program should I do this in? . . . User D A little more help Post 5 . . . You would simply do it this way: . . . Figure 1: An extract from a real-world thread individual threads, paving the way for users to spell out the general nature of their support need in their queries. The main contributions of this pa- per are: (1) designing a modular thread-level class set; (2) constructing and publishing an annotated dataset; and (3) performing preliminary thread- level experiments over the dataset. 2 Related Work There is very little work that is specifically tar- geted at …
Content truncated to 500 words
Labeled Text
This research focuses on improving information access over troubleshootingoriented technical user forums via threadlevel analysis. We describe a modular task formulation and novel dataset, and go on to describe a series of preliminary classification experiments over the data. We find that a class composition strategy achieves the best results, surpassing multiclass classification approaches. 1
Publication Details
- Paper ID:
- 28647
- Year:
- 2010
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan