Thomas Malthus once predicted that population growth would outpace the food supply  . While this hasn’t happened yet, it’s no secret that the world is well on its way to meeting the 9 billion people anticipated to be living by 2050  . While the amount of food that will have to be produced by then may be susceptible to variations―the Food and Agriculture Organization of the United Nations estimates a 60% increase―it is unquestionable that it will rise  . Those of us in the agriculture sector found ourselves asking: Will agribusiness be able to support the necessity for increased food production?
Concomitant to such a questioning scenario, agriculture has drawn peculiar attention from the general public in recent years  . Spikes in food prices such as those occurring during the 2007-2008 world food price crisis, food safety concerns such as disease outbreaks, food contaminations, environmental issues, and many others, have all raised awareness among general society  . A population, which has shifted away from rural areas and food production techniques, is now concerned with how and where food is produced.
Thanks to the internet, there is a wealth of information readily available to consumers who are now able to monitor production actions across the globe and are more conscious and exigent in decision-making. Alternatively, the outlet of social media is further catapulting information to the fingertips of consumers  .
2. The GLIMPSE Framework
The use of technology and the internet is ever increasing throughout the world and the agribusiness industry is no different. Still, the sector wrestled with what consumers really want or expect and needed a way to determine trends.
The acronym GLIMPSE was created to help the agribusiness community determine the obstacles it faces  . The original research (published in 2012) was conducted to determine the GLIMPSE framework. It was, however, more thoroughly revisited in 2015 in an effort to determine its efficacy over time.
During the second study, the researchers completed a two part analysis. Phase one was a series of interviews with 58 members of the agribusiness community. The group ranged from academic experts to industry leaders and they were asked to discuss the concerns and obstacles facing the agribusiness community.
Taking this collected data, the researchers then conducted a survey of 527 agribusiness professionals. These answers were culled down and found to follow similar concerns as those posed by the interview phase. Ultimately, it was found that for a second time, the acronym GLIMPSE resembled the primary obstacles the sector faced, but with a few changes (Figure 1 and Figure 2).
3. The Inclusion of People
The most obvious change in the revised GLIMPSE is that it now more clearly represents people. This is most obvious as it has been identified as its own category, but several of the other categories have also been altered to show the reflection of people in the form of consumers. For example, “Markets” has now been labeled “Consumer Markets” and “Losses in the food and ingredient supply chain” was adjusted to simply “Losses”, to reflect losses at the consumption level, as well as retail and production levels.
Because people have now been identified as an integral part the food chain and thus agriculture itself, it stands to reason that they should be included in the research as well. Given the advancements and spread of the internet and social media in recent years, it was considered relevant to analyze the content of posts published in these vehicles as a proxy of general public opinion. The purpose was to identify and evaluate discussions about the challenges of agribusiness and possibly draw connections to the topics previously categorized. Basically, does public opinion, represented here by social media, reflect the same obstacles and concerns as formerly identified in the interviews with academic experts and industry professionals?
4. GLIMPSE in Social Media
4.1. Crimson Hexagon
Knowing how extensive the amount of data collected could be, it became the objective to evaluate trends and patterns across the data rather than accurately measuring and classifying each and every post obtained from the sources. Therefore the analysis was mostly done based on frequency of particular words and recurrence of topics automatically classified by an artificial intelligence device known as crimson hexagon. It is a licensed commercial application that stores and searches social media content, and allows users to customize categories and analyze results.
Figure 1. The original GLIMPSE (2012)  .
Figure 2. Revised GLIMPSE (2015).
The sources of social media content analyzed included Twitter, Facebook, blogs, forums and others. The data analyzed had been posted during a three year period, from July 10th, 2012 to July 9th, 2015. Over one million social media posts were analyzed spanning this timeframe.
Upon the manual categorization of smaller samples, the system aggregates the remaining data based on similarities between the content and determined by an intrinsic algorithm. In this study, over 350 posts were manually classified according to criteria (Table 2) that followed the previously determined categories known as GLIMPSE. Any business could utilize this same tool and scour the internet for insight as to where their particular industry is trending and how to prepare for future consumer expectations.
4.3. Social Media Content Analysis
The application retrieved 1,395,652 posts meeting the search criteria. The majority of posts were published in blogs and forums. Facebook and Twitter contained the next highest level of posts, and the rest were found in accessory-type social media platforms categorized here as “Other.”
The tool enabled researchers to determine the most frequent words that could be linked to one of the GLIMPSE framework categories (Figure 3). Words such as “wa- ter”, “government”, and “health” can be easily associated to GLIMPSE categories previously described such as Environment, Government & Policies and Consumer Markets, respectively. This corroborates the comprehensiveness of the framework, but also determines that public perception as determined through social media is represented by these factors as well.
Table 1. Social media content search criteria.
Table 2. Criteria used for each category in the social media analysis (Phase 3).
Figure 3. Word frequency among posts [07/10/2012 to 07/09/2015].
Another way of analyzing the data is through clusters of words. In this analysis, the relationships of words that frequently appear together in posts are represented by interconnected bubbles. When observing these clusters (Figure 4), researchers were able to easily identify GLIMPSE categories in several of them.
When breaking down the data into different periods within the three years of content, word clouds were used to identify slight differences in trends or patterns across time. More words related to Environment and Consumer Markets categories are identified in the word cloud from 2014 to 2015, while relatively more words related to Government & Policies and Science & Innovation can be identified in the 2012-2013 word cloud (Figure 5).
When the data is segmented according to the source in which they were posted, some variations in the content can also be noted (Figure 6). Given that most of the overall
Figure 4. Cluster [07/10/2012 to 07/09/2015; sample of 10,000 posts].
data were found in blogs, the word cloud from these sources is more representative of the overall word cloud presented before. In the word clouds of Twitter and Facebook content words related to Government & Policies and Science & Innovation are hardly observed, while those related to Consumer Markets and People are predominant. It is important to note that the present analysis does not take into consideration the number of views or engagements (likes, shares, etc.) of posts, but only their content.
By observing the word clouds from each of the categories, correlation between the most frequent words and category theme can be observed. This demonstrates that the application did a fairly satisfactory job categorizing the posts. Nonetheless, some words are recurrently shown in different word clouds. The researchers believe this shows inter-relationship between GLIMPSE categories.
More importantly than the breakdown over the period is how this breakdown changed over the time or how the trend and pattern changed over time. These changes in pattern over time demonstrate changes in how people perceive the issue. Greater amount of posts related to People and Science & Innovation categories were observed in more recent posts (Figure 7).
5. Social Media Analysis Conclusions & Potential
The researchers of this subject found the social media analysis supported the findings and conclusions obtained in the previous analysis. While this was of course good news, it became increasingly evident just how beneficial this type of analysis could be for any business, government entity or policy maker, NGO, or company looking to gain perspective into the consumer mindset. The content collected from social media was top of mind to consumers; it was unprompted and completely clear of any bias from the part of the researchers.
While this particular research used Crimson Hexagon, there are other platforms available that will analyze across a wide array of information, allowing for easier deci-
Figure 7. Breakdown (%) of challenges categories in 2012/13 and 2013/14.
phering of the data. With the onset of big data, there is only to be more gain in evaluating data of this nature. As more and more consumers take their discussions, perceptions, interests, kudos or complaints to the internet, the vast amounts of data available for study are ever increasing. The information is readily available, it is up to the business world to lend a virtual ear toward social media and hear it.
A special thanks to Luiz Roberto Sodre for his dedication of time and energy regarding the social media content analysis.
 Lieberman, M. (2014) Visualizing Big Data: Social Network Analysis.
 Adami, E. (2013) National Centre for Research Methods Working Paper: A Social Semiotic Multimodal Analysis Framework for Website Interactivity.
 The Use of Social Media for Research and Analysis: A Feasibility Study.
 Connolly, A.J. and Phillips-Connolly, K. (2012) Can Agribusiness Feed 3 Billion New People...and Save the Planet? A GLIMPSE into the Future. International Food and Agribusiness Management Review, 15, 139-152.