Public opinion on OER and MOOC: A sentiment analysis of Twitter data

Abstract

The Open Educational Resources (OER) movement has gained significant momentum recently as a global effort culminating in the 2012 Paris OER declaration. However, the purist definition of OER has blurred since then, morphing into Massive Open Online Courses (MOOC). Even though OER is a significant part of the MOOC movement, it might not be a defining one. However, this has not yet been fully verified with respect to the opinion of the general public, who are the main stakeholders of both the movements. This paper attempts to explore the public opinion and perceptions regarding OER, MOOC and their complementary roles. A text mining approach is used to analyse raw Twitter data in the domains of OER and MOOC within a span of 12 months. Sentiment analysis is applied to the data to understand how public perceptions have changed during this period. The major contribution of my paper is a chronological view of public opinion on OER and MOOC, post-Paris OER declaration.

Full text

Public opinion on OER and MOOC:
A sentiment analysis of Twitter data
Ishan Sudeera Abeywardena
Wawasan Open University
Malaysia
Abstract: The Open Educational Resources (OER) movement has gained
significant momentum recently as a global effort culminating in the 2012
Paris OER declaration. However, the purist definition of OER has blurred
since then, morphing into Massive Open Online Courses (MOOC). Even
though OER is a significant part of the MOOC movement, it might not be
a defining one. However, this has not yet been fully verified with respect
to the opinion of the general public, who are the main stakeholders of
both the movements. This paper attempts to explore the public opinion
and perceptions regarding OER, MOOC and their complementary roles. A
text mining approach is used to analyse raw Twitter data in the domains of
OER and MOOC within a span of 12 months. Sentiment analysis is applied
to the data to understand how public perceptions have changed during
this period. The major contribution of my paper is a chronological view of
public opinion on OER and MOOC, post-Paris OER declaration.
Keywords: open educational resources, OER, MOOC, text mining, opinion
mining, sentiment analysis, open education
1 Introduction
As a result of the new drive towards accessible and open information,
Open Educational Resources (OER) have taken centre stage after being first
adopted in a UNESCO forum in 2002. An early definition of OER is ‘web-
based materials, offered freely and openly for use and re-use in teaching,
21
ICOFE Book.indd 296 2014/10/13 下午 04:01:53

Public opinion on OER and MOOC: A sentiment analysis of Twitter data
297
learning and research’ (Joyce, 2007, p. 1). The Paris OER Declaration
(UNESCO, 2012, p. 1) provides a more comprehensive definition:
teaching, learning and research materials in any medium, digital or
otherwise, that reside in the public domain or have been released
under an open license that permits no-cost access, use, adaptation and
redistribution by others with no or limited restrictions. Open licensing
is built within the existing framework of intellectual property rights
as defined by relevant international conventions and respects the
authorship of the work.
Arguably, the months since the Paris OER Declaration are the most crucial
in the future direction of the OER movement. Many new OER initiatives
have blossomed since addressing the 10 key recommendations made to
policymakers. On the other end of the spectrum, Massive Open Online
Courses (MOOC) have gained momentum as an innovative way of
increasing access and equity in education. Daniel (2012) argues that the
concept of MOOC is also constantly evolving, trying to define itself within
the open education movement. In his article he quotes the Wikipedia
definition of MOOC: ‘a MOOC is a type of online course aimed at large-
scale participation and open access via the Web’ (Daniel, 2012, p. 3). There
is constant academic debate currently taking place with respect to the
origins of the concept of MOOC. However, MIT’s MITx programme, which
has now morphed into edX, is accredited as the first implementation of
MOOC. Since then, Stanford University has launched its MOOC platform,
Udacity, which doubles as a commercial entity providing services to new
MOOC startups. Coursera is another example of a commercial start-up
which claims to have, at the end of 2012, over 1.4 million learners enrolled
in more than 200 courses offered by 33 partner institutions (Lewin, 2012;
DeSantis, 2012). All of these comprehensive courses are openly and freely
available to global learners, potentially bridging the knowledge divide.
McAuley, Stewart, Siemens, and Cormier (2010, p. 6) claim that
The large scale of the community, from several hundred to several
thousand participants, maximizes the possibility that the ‘long tail’
effect will enable someone with even the most esoteric interests within
the overall focus of the MOOC to find people with whom to share and
collaborate.
ICOFE Book.indd 297 2014/10/13 下午 04:01:53

Ishan Sudeera Abeywardena
298
Thus, MOOC have wide implications on how education will be perceived
in the future. Although OER play a major role in the delivery of MOOC, it
might not be a defining one. As such, the two camps are partially divided
when it comes to which is the best way forward. Affirming this, Daniel
(2012, p. 13) argues that
...what MOOCs will not do is address the challenge of expanding
higher education in the developing world. It may encourage
universities there, both public and private, to develop online learning
more deliberately, and OER from MOOC courses may find their
way, alongside OER from other sources, into the teaching of local
institutions.
This division results in confusion among the stakeholders, because they are
now faced with a difficult choice in organisational, institutional or national
policy. In an attempt to identify the extent of this dilemma, this paper
looks at the sentiment of the public (stakeholders) with respect to OER and
MOOC.
‘Sentiment’, as described by Pang and Lee (2008, p. 1), is the automatic
analysis of evaluative text and tracking of the predictive judgments.
Therefore, the activity of sentiment analysis can be best explained as ‘the
computational treatment of opinion, sentiment, and subjectivity in text’
(Pang & Lee, 2008, p. 1). Relating this to social media, Zabin and Jefferies
(2008, p. 327) argue that
With the explosion of Web 2.0 platforms such as blogs, discussion
forums, peer-to-peer networks, and various other types of social media
. . . consumers have at their disposal a soapbox of unprecedented
reach and power by which to share their brand experiences and
opinions, positive or negative, regarding any product or service. As
major companies are increasingly coming to realize, these consumer
voices can wield enormous influence in shaping the opinions of
other consumers — and, ultimately, their brand loyalties, their
purchase decisions, and their own brand advocacy. . . . Companies can
respond to the consumer insights they generate through social media
monitoring and analysis by modifying their marketing messages, brand
positioning, product development, and other activities accordingly.
ICOFE Book.indd 298 2014/10/13 下午 04:01:53

Public opinion on OER and MOOC: A sentiment analysis of Twitter data
299
To achieve the objective of identifying the public opinion on OER
and MOOC, a text mining approach is used to extract data from social
media for sentiment analysis. The major contribution of this paper is a
chronological view of public opinion on OER and MOOC for a span of 12
months, post-Paris OER declaration. Through this view, a roadmap can be
identified for future research and development based on public demand.
This is the major advantage of the preliminary findings presented.
The remainder of the paper is divided into methodology, results, discussion
and conclusion.
2 Methodology
At present, Twitter (twitter.com) is one of the largest social networking
platforms in existence, used by over 230 million monthly active users
(Twitter, 2013). The platform allows users to share concise (140 characters)
posts in the form of microblogs called tweets. These tweets are seen by
selected individuals, groups or publically by the whole Twitter network,
depending on the social status set. Following a tweet, the social interaction
takes place in the form of response tweets or as retweets, whereby the
initial tweet is shared virally. This rich interaction results in a large number
of tweets being generated on a particular topic.
It is observed that the OER and MOOC communities actively use social
media such as Twitter for teaching and learning purposes (McAuley et al.,
2010). Furthermore, due to the unpoliced and liberal nature of this social
networking platform, frank opinions are shared by users on the benefits
as well as the shortcomings of both ideologies. As such, Twitter was found
to be the ideal data source to analyse public opinion regarding OER and
MOOC.
The extraction of tweets was done manually using the native search
mechanism of Twitter (twitter.com/search-home). The search terms ‘OER’
and ‘MOOC’ were used in the search. Only the abbreviated forms were
used because I had identified empirically that the terms ‘Open Educational
Resources’ and ‘Massive Open Online Courses’ were seldom used in tweets.
This is mainly due to the 140 character restriction of a tweet. Only the
‘Top’ tweets were extracted, because these are the best matches for a given
ICOFE Book.indd 299 2014/10/13 下午 04:01:53

Ishan Sudeera Abeywardena
300
search query. Furthermore, the use of the Top tweets acted as a first-level
cleansing of the dataset. The raw data were organised using the MS Excel
2010 spreadsheet application. A second-level cleansing of the dataset was
conducted manually to remove partial, irrelevant and repeated tweets.
Only distinct tweets were used. All user details were removed to ensure
anonymity and unbiased analysis. The details of the final dataset are
provided in Table 1. Due to the large volume of tweets on ‘MOOC’ it was
found challenging to manually extract all the tweets beyond 6 months.
Furthermore, when considering the number of distinct tweets, those on
‘MOOC’ for six months were considered to be a sufficient sample size in
comparison to the distinct tweets on ‘OER’ for 12 months.
Table 1 The Twitter dataset used in the study consisting of tweets which include the
terms ‘OER’ and ‘MOOC’
Search query Tweet history Time No. of distinct
tweets
‘OER’ 1/11/2012 – 31/10/2013 12 months 1209
‘MOOC’ 1/05/2013 – 31/10/2013 6 months 2823
The sentiment analysis was done using the Semantria (semantria.com)
software application, which comes in the form of a plug-in for the MS
Excel spreadsheet application. In preparation for analysis, an identity
column was added to the dataset to enable the analysis of individual tweets
with respect to sentiment. A basic sentiment analysis was conducted on the
dataset using the Semantria plug-in. The plug-in uses a cloud-based corpus
of words tagged with sentiments to analyse the dataset. Through statistical
inferences, each tweet is tagged with a numerical sentiment value ranging
from -1.5 to +1.5 and a polarity of (i) negative, (ii) neutral, or (iii) positive.
The positivity increases with the sentiment value. The sentiment data is
then reorganised according to individual months of the tweet history, to
identify public opinion on a particular topic over a given time.
3 Results
The analysis of the data looks at two major aspects: (i) the number of
tweets over a given time, and (ii) the public sentiment over a given time.
ICOFE Book.indd 300 2014/10/13 下午 04:01:54

Public opinion on OER and MOOC: A sentiment analysis of Twitter data
301
Figure 1 and Figure 2 show the number of tweets on ‘OER’ and ‘MOOC’
respectively. Figure 3 shows the public opinion on ‘OER’ over a 12-month
span, and Figure 4 shows the public opinion on ‘MOOC’ over a 6-month
span.
Figure 1 Total tweets on ‘OER’ for 12 months from November 2012 to October 2013.
Figure 2 Total tweets on ‘MOOC’ for 6 months from May 2013 to October 2013
ICOFE Book.indd 301 2014/10/13 下午 04:01:54

Ishan Sudeera Abeywardena
302
Figure 3 Public opinion on ‘OER’ over 12 months
Figure 4 Public opinion on ‘MOOC’ over 6 months
ICOFE Book.indd 302 2014/10/13 下午 04:01:54

Public opinion on OER and MOOC: A sentiment analysis of Twitter data
303
Figure 5 provides a comparison between the numbers of tweets on ‘OER’
vs. ‘MOOC’ for a 6-month span between May and October 2013. Figure 6,
Figure 7 and Figure 8 depict the change in positive, neutral and negative
public opinion respectively on ‘OER’ vs. ‘MOOC’ for the same time.
Figure 5 Comparison between the numbers of tweets on ‘OER’ vs. ‘MOOC’ for 6
months between May and October 2013
Figure 6 Change in positive public opinion on ‘OER’ vs. ‘MOOC’ for 6 months between
May and October 2013.
ICOFE Book.indd 303 2014/10/13 下午 04:01:54

Ishan Sudeera Abeywardena
304
Figure 7 Change in public neutral opinion on ‘OER’ vs. ‘MOOC’ for 6 months between
May and October 2013
Figure 8 Change in negative public neutral opinion on ‘OER’ vs. ‘MOOC’ for 6 months
between May and October 2013
ICOFE Book.indd 304 2014/10/13 下午 04:01:54

Public opinion on OER and MOOC: A sentiment analysis of Twitter data
305
4 Discussion
With reference to Figure 5, it can be seen that there is relatively more
discussion taking place on the topic of MOOC than on OER. As shown in
Table 1, the total number of distinct tweets on OER for 12 months is only
42.8% of the total number of distinct tweets on MOOC within 6 months.
Furthermore, Figure 1 and Figure 2 suggest that the interest in OER is on a
downward trend, whereas the interest in MOOC is on the rise. There may
be several factors contributing to this trend. Among them, the novelty of
MOOC, the involvement of the private sector, the brand names associated
with the recent MOOC delivered, the keen interest of highly reputed
conventional institutions in the concept and the large marketing budgets
could be key influences.
Although the interest in OER seems to be declining, Figure 3 suggests
that the public opinion on OER remains largely positive. By comparing
the averages of the positive values shown in Figure 3 and Figure 4, it can
be seen that the average positive public opinion on OER is 31.19% in
comparison to 25.16% on MOOC.
Considering the span of 6 months from May to October 2013, the change
in positive public opinion on OER and MOOC is highlighted in Figure 6.
Even though the number of tweets on MOOC is considerably larger than
on OER, suggesting more interest in the former, public opinion on MOOC
seems to remain unchanged with respect to the positive effect it has. In
contrast, the stakeholders seem to feel more positive about OER and the
benefits it has. Furthermore, this positivity towards OER seems to be on an
upward trend.
When considering the negative public opinion on OER and MOOC, Figure
8 suggests that the negativity towards both concepts is lessening. However,
the negativity towards OER is slightly less than it is towards MOOC. The
more interesting trend is the change in public neutrality towards OER and
MOOC. Figure 7 shows only a slight change in neutrality towards MOOC,
whereas there is a decline in public neutrality towards OER. This suggests
that the public is now making informed opinions regarding OER, whereas
caution and scepticism still surround the relatively new concept of MOOC.
ICOFE Book.indd 305 2014/10/13 下午 04:01:54

Ishan Sudeera Abeywardena
306
The strength of this study is that it takes an objective look at both MOOC
and OER from the public’s perspective. In contrast to an instrument-
driven survey which would restrict the expression of opinion by the
public to a certain reference frame, the use of sentiment analysis takes into
consideration all that the public has said and the relative connotations.
Although a considerably large number of tweets has been analysed, it can
be argued that the sample size is not sufficient to generalize about any
trends. This is one of the weaknesses of the study. However, the advantage
is that this model can be replicated using a larger sample size when the
data are available in the future.
5 Conclusion
The debate on Open Educational Resources (OER) vs. Massive Open Online
Courses (MOOC) has been raging for the past few of years. Advocates
of the two camps have been bombarding the public with the benefits of
these ideologies. Given that both OER and MOOC are pervasive forces
which can change the education landscape of the future, the general
public (stakeholders) are in a state of confusion with respect to adopting a
particular ideology for their institution, organisation or nation. This study
looks at OER and MOOC from the public’s perspective. It uses sentiment
analysis of Twitter data to approximate the public opinion on OER and
MOOC according to the benefits they promise.
The analysis of the Twitter data suggests that there is growing interest in
MOOC in comparison to OER. However, when considering public opinion
on MOOC, it is apparent that the stakeholders haven’t yet formed strong
opinions on MOOC, due to it being a relatively recent phenomenon.
In contrast, the positivity towards OER is on an upward trend. The
neutrality towards OER has also decreased over 12 months, post-Paris
OER declaration, suggesting that the public are slowly but surely forming
informed opinions about the use of OER.
The contribution of this paper is a chronological view of how the interest
and public opinion have changed with respect to OER and MOOC
following the 2012 Paris OER declaration. Three main trends can be
identified which have to be taken into consideration by the stakeholders
ICOFE Book.indd 306 2014/10/13 下午 04:01:54

Public opinion on OER and MOOC: A sentiment analysis of Twitter data
307
when making a policy decision: (i) there is an increasing amount of interest
on MOOC, (ii) the public still hasn’t formed strong opinions regarding
MOOC due to the novelty of the ideology, and (iii) the positivity towards
OER is growing.
This paper discusses the preliminary analysis of the data. It is my intention
to further probe the data and identify the reasons behind the trends
identified in this paper. The next step is to potentially pinpoint certain
phenomena which would have a positive or negative impact on the public
opinion with respect to MOOC and OER.
References
Abeywardena, I. S. (2014). Public opinion on OER and MOOC: A sentiment
analysis of Twitter data. International Conference on Open and Flexible
Education. Hong Kong SAR: Open University of Hong Kong.
Daniel, J. (2012). Making sense of MOOCs: Musings in a maze of myth,
paradox and possibility. Journal of Interactive Media in Education, 3.
DeSantis, N. (2012). After leadership crisis fuelled by distance-ed debate,
UVa will put free classes online. Chronicle of Higher Education.
Joyce, A. (2007). OECD study of OER: Forum report. UNESCO.
Lewin, T. (2012). Education site expands slate of universities and courses.
New York Times.
McAuley, A., Stewart, B., Siemens, G., & Cormier, D. (2010). The MOOC
model for digital practice.
Pang, B., & Lee, L. (2008). Opinion mining and sentiment analysis.
Foundations and Trends in Information Retrieval, 2(1–2), 1–135.
Twitter. (2013). About Twitter, Inc. Retrieved October 3, 2014 from https://
about.twitter.com/company.
UNESCO. (2012). 2012 Paris oer declaration. Retrieved October 3, 2014
from http://www.unesco.org/new/fileadmin/MULTIMEDIA/HQ/CI/
WPFD2009/English_Declaration.html.
Zabin, J., & Jefferies, A. (2008). Social media monitoring and analysis:
Generating consumer insights from online conversation. Aberdeen
Group Benchmark Report, 37(9).
ICOFE Book.indd 307 2014/10/13 下午 04:01:54