Essay: Sentiment Analysis On XYZ Social Network - Exploring Social Media Platforms.

Related Research

The increased popularity and acceptability of social media platforms has made them an important source of data and information. This data aspect of the social media platforms has attracted a lot of interest from individuals and firms that want to carryout research relating to their brands (Farris & Neil, 2010; Anthony & Johnson, 2008). The main aim being to get an understanding of the needs of the clients or prospective clients.

Another interesting field of research on data generated from social media platforms is sentimental analysis. This kind of research enables the social media platforms to gauge the attitudes of individual users of their platform. This could be randomly or in response to a given preset condition. Such a research enables the social media platform to better understand the users of the site.

XYZ Social Network is a social media site that enables the online interaction of people. This research aims at observing and analyzing the nature of the posts of users on the XYZ site with the aim of establishing whether the social media platform can alter the emotions of its users.

A number of research works have been done on determining the nature of the emotions of the users of social media platforms. This has been done through two main methods; Observation only method, where the attitude of a social media post is randomly assessed and Intervention-Observation method, where the feed of the users of a social media platform is altered and the attitude of their posts thereafter observed.

In (Apoorv, et al., 2011), only observations are made. The Twitter social media platform is regarded as a form of a microblog site. The attitudes of the tweets of a selected sample are analyzed in this research. The aim of the research is to determine whether the posts were positive, negative or neutral.

Intervention-Observation method was used for a controversial research by Facebook. (Kramer, et al., 2014) Applies an intervention technique in which the feed received by a sample of users on Facebook is tailored to be either positive or negative. The posts of the sample of users are then analyzed to determine their attitude. The purpose of the research was to establish whether Facebook could successfully alter the emotions of its users. The research work, (Kramer, et al., 2014), was however faced with ethical issues pertaining to the participants’ consent.

Advances in sentimental analysis have been attributed to the growth of online platforms for blogging, microblogging and reviewing (Rajkumar, et al., 2014). Observation only method is also applied for the research into reviews for products and services from Twitter in (Rajkumar, et al., 2014). The research primarily focuses on analyzing the attitude from the reviews given for various products on Twitter.

Research Methodology

This research, unlike (Apoorv, et al., 2011) and (Rajkumar, et al., 2014), aims at using the Intervention-Observation method to evaluate the emotions of users on the XYZ Social Network platform. However, different from what was done by (Kramer, et al., 2014), this research will factor in the ethical considerations for research in order to provide reliable and acceptable findings.

Paradigm

Individuals’ perceptions change depending on the condition or situation they find themselves in. A person cannot be said with certainty to be of a given opinion or to be having a particular general mood. All these rely on the environment in which a person is, the context of the situation that the person finds themselves in. Therefore, the perception and reality of an individual is at a constant point of creation or generation depending on the situation they are in.

Hence, the perceptions of individuals need to be observed and consideration given to the prevailing circumstances the individuals find themselves in. This will in turn help in better understanding of the thinking and actions and factors associated with the thinking and actions of the individual.

Methodology

A qualitative research approach will be applied for the research. The type of qualitative approach that will be used in the research is the Content Analysis Approach.

Content Analysis refers to a qualitative research technique where the contents of the response given by participants are analyzed and understanding derived from the analysis (Creswell, 2014).

In this research, the contents of the posts of the participants will be put through analysis and deductions made as to whether the posts are positive, negative or indifferent in nature. The deductions will then be compared with the nature of the feed that the participant was exposed to, in order to determine whether the XYZ Social Network Company can alter the emotions of the users of its site.

Sampling

This research will apply the Quota Sampling Method to determine the composition of the sample of users of XYZ Social Network to be approached (through messages) for invite to participate in the research.

The Quota Sampling Method is a non-probability sampling technique where a desired attribute is identified in the population and the proportion for the presence of the attribute is noted and used in the sample selection (Dillman, et al., 2009). The proportion for the presence of the attribute is then taken as the proportion for the presence of the attribute in the sample size selected (Freedman, 2009). The Quota Sampling technique helps in making the selected sample more representative of the entire population with respect to the attribute of interest.

Text Mining and Sentiment Analysis

In this research, we will consider the proportions of the gender and age attributes of the entire population of the XYZ Social Network users. The proportions for these two attributes will then act as a guide for the proportions of the same attributes in the sample selected for the research.

In the determination of the sample size for this research, the Cochran’s Sample Size Formula will be used. (Cochran, 1977) Uses the following parameters and variables for computation of the sample size; standard margin of error (usually 1%), standard confidence interval (usually 95%) and the population size. The (Cochran, 1977) formula for sample size determination is given as below;

Z = from Z tables,

a = attribute under consideration,

p = proportion of the attribute under consideration in the population,

q = p-1,

e = standard margin of error.

The data for this research will be collected using the text mining technique. Text mining is a data mining tool in which information in the form of texts is collected from sources such as social media sites and stored as matrices called text-document matrices (Galit, et al., 2018).

The text mining process will collect the posts of the participants from the XYZ Social Network platform and rearrange them in a manner that will enable the application of the analytical tools of data mining. The algorithm for text mining makes use of the API (Application Programming Interface) of the Social Network site, in this case the XYZ Social Network site. The API grants the algorithm the permission required to collect the posts from the XYZ Social Network Site. The algorithm also allows for the selection of the specific accounts and periods for the collection of the posts.

The data analysis technique that will be applied in this research will be the Sentimental Analysis. This is a form of Content Analysis research approach that is used to gauge the attitude of responses given to a question or series of questions (Huifeng, et al., 2009).

Sentimental Analysis is a data mining technique that makes use of the text mining tools for collection of text data and applying Biometrics, Computational Linguistics and/or Natural Language Processing (NLP) for the analysis (Taboada & Brooke, 2011).

In this research the type of Sentimental Analysis that will be used is the Polarity Test Sentimental Analysis. This type of Sentimental Analysis evaluates the mood or attitude of a given statement as positive, negative or indifferent (Huifeng, et al., 2009).

Validity and Reliability of the Research

For the case of this study, interest will be in classifying the posts of the participants in the research as positive, negative or indifferent. After classification, the outcome will then be compared with the nature of the feed that the given participant was exposed to (that is, positive or negative feed).

This comparison will then be evaluated to give an indication as to whether the XYZ Social Network Site is able to alter the emotions of the users of the site.

To evaluate the soundness of this research project, the following aspects will be considered:

• Validity
• Reliability
• Credibility
• Trustworthiness

Validity can be described as the extent to which the research instruments in a research are effective in the measurement of the aspects that the research aims at measuring (Mohajan, 2017). The validity gauges the appropriateness of a research instrument for a particular research.

In this research we apply Content Analysis as the research instrument. Content Analysis as a research design aims at observing the contents of a respondent (be it audio, visual or text) and deriving conclusions with regards to a particular topic (Creswell, 2014).

This research mainly aims at determining whether we have positive, negative or indifferent posts from users of the XYZ Social Network depending on the nature of the feeds that they are exposed to. This makes the Content Analysis design appropriate for collecting the information on the nature of the content through the application of Sentimental Analysis, thereby making the research valid.

Reliability is the measure of the extent to which a research is repeatable, precise and consistent (Chakrabartty, 2013). The reliability of a research checks on whether a given research is dependable both in terms of the process and the results.

In this research Content Analysis will be carried out repeatedly for the posts of the participants on the XYZ Social Network. This will enable the consistency of the results to be measured. For instance, if the first post of a participant after the start of the research is classified as negative through the sentimental analysis, the second and third posts after the start of the research will also be analyzed to check whether they will also be negative.

Credibility can be defined as the extent to which the findings of a research relates to the reality of the research matter (Lisa, 2017). The credibility of the research addresses the question of bias in the data collection, analysis and conclusion of a research project.

In this study, we apply member-checking technique to ensure the credibility of the research project. In the member-checking technique, regular updates of research progress and feedbacks are shared with the participants (Tanggaard, 2008). Through this communication channel, participants are able to affirm or disregard the findings at any stage of this research. This will thus improve the credibility of the research.

Another technique that will be applied in order to improve the credibility of this research is repeated sampling. Data will be collected from the same sample of participants repeatedly and Content Analysis done and comparison made with findings from previous samples.

Trustworthiness of a research project is the measure of the level transferability and confirmability of a research (Marshall & Rossman, 2011). Trustworthiness gauges the extent to which the findings of a research can be applied for similar cases. It also concerns the level of bias on the part of the participants, whether their responses were truly theirs (Loh, 2013).

In this research, the focus is in determining whether the emotions of the users of XYZ Social Network can be altered depending on the feed they are exposed to. The findings from this research can easily apply for the cases of other social networking platforms, since the platforms operate in the same basic manner.

The confirmability, in the case of this research, is guaranteed through the consent forms that are made available for the intended participants in the research. The consent forms provide the participants with all the information that they need regarding the research. This ensures that the users of XYZ Social Network that agree to participate, do it out of free will without any coercion hence removing the aspect of bias in the research.

• In instances where a participant writes a sarcastic post, it would be difficult to establish the overall mood (positive, negative or indifferent) of the post using sentimental analysis.
• Participants may also write posts that can be deemed as evenly balanced between positive and negative mood, this makes the analysis of the overall mood impossible. This could however be classified as indifferent.
• The participant consent-based approach in research projects exposes the study to sample size issues. The number of individuals that agree to participate may fall short of the intended sample size for the study.  

In order to avoid the research being termed as unethical, the ethical issues below will be considered and addressed in this research project.

It is crucial to have consent from the participant in any research study. This is provided in the form of a consent form to be signed by the intended participants. The participants’ consent, especially, informed participants’ consent gives the research study validity, with respect to the source of the data used in the research. In addition to this, the informed participants’ consent makes the risks and the benefits of the research study aware to the intended participants (Marshall, 2007).  

In order to address the issue of participants’ consent, this research will start by sending messages to the sample of users of the XYZ social network targeted for this research. This message will invite the individuals to participate in the research and also provide all the relevant information to ensure that they are adequately informed about the study.

The message will be accompanied with a link to an online consent form with the terms and condition for participation in the research as well as consent section. In the consent section the participant will be asked to confirm to have read the terms and condition for the participation in the study and whether they have agreed to take part in the research.

Individuals should willingly take part in a research without any aspect of coercion or intimidation. Voluntary participation improves the reliability of the research as a reference point since the data is considered as unbiased (Igor & Lindsey, 2016).

Individuals taking part in a research study should also be given the liberty to opt out of the study if they so wished. Participants should have the option to exclude themselves from the study at any point of the study if they felt no longer interested or willing to continue taking part (Biggs & August, 2013).

The voluntary participation and right of withdrawal will be included in the terms and condition as part of the online consent form. The intended participants will be made aware that participation in the research is entirely voluntary and that they have the right to withdraw their participation in the research at any stage of the research.

In any research, it is important to guarantee the protection of the identity of the participants. The participants’ identity protection becomes very vital in instances where the research touches on a socially or politically sensitive issue (Pierre, 2011). However, regardless of whether the research touches on a sensitive issue or not, the participants should be guaranteed that their identities will be protected. Also important to guarantee is the protection of the records of the outcomes of the research from every participant.

In this research, the identities of the respondents and their respective research outcomes will not form part of the eventual report publication after the data collection and analysis has been done. Although this is the standard practice, consideration could be given to participants that would wish to be acknowledged for their contribution to the research. This would be catered for by including such a provision in the online consent form asking the participants whether they would like to be acknowledged for their contribution to the research.

Hacking and illegal access to data has been a point of concern with the advancement and increased reliance on technology. Research data has been susceptible to illegal access and hacking (O'Neil, 2011). The data can be used to make public the identities and analysis outcomes for the respective participants. The data may also be used for the preparation of other studies that it was not intended for. This will constitute a breach of the terms and conditions in the participants’ consent form thereby exposing the research project to legal action.

In this research, relevant data from the research should be stored in the servers of the XYZ Social Network Company. This will ensure that the data will be secure alongside the account information of the users of the XYZ Social Network. This will also be a cost effective approach to secure storage of research data as opposed to acquiring the services of a third party for the storage.

The participants should have access to the progress of the research project. This is by virtue of their participation and by extension, their contribution to the gathered data. Regular feedbacks and updates on the progress of the research project at all stages gives the participants a sense of ownership of the project (Kimmelman, et al., 2011).

With prompt feedback to queries and regular updates on the progress, the research gains acceptability with the participants. This is especially in occasions where a participant hesitantly agreed to participate and thus there is a high chance of them withdrawing from the research at some stage. The feedbacks and updates gains their trust and confidence in the research project.

Conclusions

The following conclusions can be drawn from the research design report:

• Due to the nature of the data to be collected (that is; in form of posts), text mining technique is the most effective and appropriate method of data collection.
• Content Analysis, and more specifically Sentimental Analysis will be a reliable data analysis tool for the research.
• The various measures taken for the ETHICAL CONSIDERATIONSin the research design will prevent the research from being termed as unethical.

References

Anthony, S. D. & Johnson, M. W., 2008. Innovator's Guide to Growth. 1st ed. New York: Havard Business School Press.

Apoorv, A., Xie, B., Rambow, O. & Passonneau, R. J., 2011. Sentimental Analysis of Twitter Data, New York: Columbia University.

Biggs, J. S. & August, M., 2013. Rebuilding a Research Ethics Committee. Journal of Research Administration, 44(1), pp. 1-3.

Chakrabartty, S. N., 2013. Best Split-Half and Maximum Reliability. IOSR Journal of Research and Method in Education, 3(1), pp. 1-8.

Cochran, W. G., 1977. Sampling Techniques. 3rd ed. New York: John Wiley & Sons.

Creswell, J. W., 2014. Research Design: Qualitative, Quantitative and Mixed Approaches. 4th ed. Michigan: SAGE Publications, Inc.

Dillman, D. A., Smyth, J. D. & Christian, L. M., 2009. Internet, Mail and Mixed-Mode Surveys: The Tailored Design Method. San Francisco: Jossey-Bass.

Farris, P. W. & Neil, B. T., 2010. Marketing Metrics. 2nd ed. New Jersey: Pearson Education.

Freedman, D. A., 2009. Statistical Models: Theory and Practice. 1st ed. London: Cambridge University Press.

Galit, S., Peter, B. C., Inbal, Y. & Nitin, P. R., 2018. Data Mining for Business Analytics. 1st ed. s.l.:John Wiley & Sons, Inc..

Huifeng, T., Tan, S. & Cheng, X., 2009. A Survey on Sentiment Detection of Reviews. Expert Systems with Applications, 36(7), pp. 10760-10773.

Igor, G. & Lindsey, M., 2016. Alternative Models of Ethical Governance: The 2016 New Brunswick-Otago Declaration on Research Ethics. New Zealand Sociology, 31(4), pp. 1-7.

Kimmelman, J., McDonald, M. & Avard, D., 2011. A Guide to the Perplexed: How to Navigate Conflicting Research Ethics Policies. Health Law Review, 19(3), pp. 1-4.

Kramer, A., Guillory, J. & Hancock, J., 2014. Experimental Evidence of Massive-scale Emotional Contagion Through Social Networks. Proceedings of the National Academy of Sciences, 111(24), pp. 8788-8790.

Lisa, M. P., 2017. A Framework for Determining Research Credibility. Crimson Publishers, 1(1), pp. 1-4.

Loh, J., 2013. Inquiry into Issues of Trustworthiness and Quality in Narrative Studies: A Perspective. The Qualitative Report, 18(65), pp. 1-15.

Marshall, C. & Rossman, G. B., 2011. Designing Qualitative Research. 5th ed. Los Angeles: SAGE Publications.

Marshall, P. A., 2007. Ethical Challenges in Study Design and Informed Consent for Health Research in Resource-Poor Settings. Special Topics in Social, Economic and Behavioral Research, 2007(5), p. 2.

Mohajan, H., 2017. Two Criteria for Good Measurements In Research: Validity and Reliability. Annals of Spiru Haret University, 17(3), pp. 58-82.

O'Neil, P., 2011. The Evolution of Research Ethics in Canada; Current Developments. Canadian Psycology, 52(3), pp. 2-9.

Pierre, D., 2011. A New Perspective on Research Ethics. Health Law Review, 19(3), pp. 1-5.

Rajkumar, S. J., Vishal, S. S. & Sachin, N. D., 2014. Sentiment Analysis of Events from Twitter Using Open Source Tool. International Journal of Computing Science and Mobile Computing, 5(4), pp. 475-485.

Taboada, M. & Brooke, J., 2011. Lexicon-based Methods for Sentiment Analysis. Computational Linguistics , 37(2), pp. 272-274.

Tanggaard, L., 2008. Objections in Research Interviewing. International Journal of Qualitative Methods, 7(3), pp. 15-29