Predicting Subjectivity Orientation of Online Forum Threads Page: 2
The following text was automatically extracted from the image on this page using optical character recognition software:
he conveys to some camera forum's search engine by issuing the following queries:
1. "How is the resolution of canon 7D?", and 2. "What is the resolution of canon
7D?". Both queries are about the resolution of canon 7D (and may look similar
at first sight) but the user's intent is different across the two queries. In the
first query, the user seeks opinions of different camera users on the resolution of
the Canon 7D camera, i.e., how different users feel about the resolution, what
are their experiences (good, bad, excellent, etc.) with Canon 7D as far as its
resolution is concerned; hence, the query is subjective. In the second query, the
user does not seek opinions but an answer to a specific question, which in this
case, is the value of the resolution and therefore the query is non-subjective.
Hence, prior knowledge of the subjectivity of threads would help in satisfying
users' information needs more effectively by taking into account the user's in-
tent in addition to the keywords in the query. In order to answer such queries
effectively, forum search engines need to identify subjective threads in online fo-
rums and differentiate them from threads providing non-subjective information.
Threads can be filtered by matching their subjectivity orientation with that of
the query or they can be ranked by combining scores of lexical relevance and
subjectivity match with the query.
Here, we address the first part of this vision; we show how to identify the sub-
jectivity of threads in an online forum with high accuracy using simple word fea-
tures. Recent works on online forum thread retrieval have taken into account the
distinctive properties of online threads such as conversational structure , and
hyperlinking patterns and non-textual metadata  to improve their retrieval.
Previous works on subjectivity analysis in social media have mainly focused on
online review sites for opinion mining and sentiment analysis [3,4,5] and on im-
proving question-answering in community QA [6,7,8,9]. In contrast, our focus is
on analyzing subjectivity in online forums using content based features.
We propose a simple and effective classification method using textual features
obtained from online forum threads to identify subjective threads of discussion.
We model the task as a binary classification of threads in one of the two classes:
subjective and non-subjective. We say a thread is subjective if its topic of dis-
cussion is subjective and non-subjective if its topic is non-subjective. We used
combinations of words and their parts-of-speech tags as features. The features
were generated from the text in: (i) the title of a thread, (ii) the title and initial
post of a thread and (iii) the entire thread. We performed experiments on two
popular online forums (Dpreview and Trip Advisor-New York forums). We used
ensemble techniques to improve learning of classifiers on unbalanced datasets
and also explored the effects of feature selection to improve the performance of
our classifiers. Our experiments show that our classifiers using textual features
produce highly accurate results with respect to F1-measure.
Our contributions are as follows. We show that simple features generated
from n-grams and parts-of-speech tags work effectively for identifying subjective
and non-subjective discussion threads in online forums. We believe that online
forum search engines can improve their ranking functions by taking into account
the subjectivity match between users' queries and threads.
Here’s what’s next.
This chapter can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Chapter.
Biyani, Prakhar; Caragea, Cornelia & Mitra, Prasenjit. Predicting Subjectivity Orientation of Online Forum Threads, chapter, March 2013; [Berlin, Germany]. (digital.library.unt.edu/ark:/67531/metadc725770/m1/2/: accessed April 23, 2018), University of North Texas Libraries, Digital Library, digital.library.unt.edu; crediting UNT College of Engineering.