Mahir Shahriar
ENGL 21007
Writing For Engineering
 
         In recent years, we have seen a massive increase in the power and use of Artificial Intelligence (AI). Around 30 years ago, AI research existed, but it was very much not making progress. This was mainly due to the lack of advancement in both hardware and software. We just did not have the materials needed to be able to make it better. However, in the recent 20 years, many advancements in computer chips, microprocessors, neural links, and other hardware and software have allowed AI research and development to flourish. We live in an age where AI is available for anyone who has access to a device. However, one flaw that AI has had in the past was emotion recognition. AI was limited in that sense, which made it difficult to be utilized in settings such as healthcare, marketing, and humanoid robot development. All of these fields require that the AI gather information about the emotions being displayed, and use them to make the best possible decision. Luckily, lots of research has gone into emotion recognition over the past couple of years, and many advancements have been made. With the topic of emotion recognition in AI in mind, I chose to analyze the lab report, “Emotion Recognition and Artificial Intelligence: A Systematic Review (2014-2023) and Research Recommendations” by Smith K. Khare, Victoria Blanes-Vidal, Esmaaeil S. Nadimi, and U. Rajendra Acharya. This lab report extensively goes over the current research done on emotion recognition in AI, while also giving insight on how future research could be done. This lab report somewhat follows the general structure of lab reports as shown in Chapter 19 of “Technical Communication” by Mike Markel but deviates from it during two of the main sections and I will be analyzing each of the elements to see how the format affects the narrative of the lab report.
Title
         The title of the lab report is fairly simple. “Emotion recognition and artificial intelligence: A systematic review (2014-2023) and research recommendations. This title is both precise and informative enough about what the lab report is about which makes it easy for researchers can see what exactly is contained in the lab report. From the title, we know that the report covers a review of past research from 2014 to 2023 on emotion recognition in artificial intelligence.
Abstract
         The abstract of a lab report summarizes the entire lab report in a short, few sentences.  It is meant to parallel the structure of the lab report. The type of abstract given in this lab report is a descriptive abstract, which only states the topics that are being covered in the report as opposed to an informative one, which covers the major findings, results, and conclusions. This type of abstract makes it so that if a researcher wanted to use this report, they would have to read through the entire report to find out the major conclusions. While normally, this would be a turn-off, it makes sense why the writers chose a descriptive abstract over an informative one. This report is more of a compilation of all the past research from the years 2014-2023. If they did an informative abstract, the abstract would have had to summarize each of the studies used in this report, which would have been lengthy.
Introduction
         The introduction of this report brings into context of why emotional recognition in the application of AI is important. It talks about the uses it has in healthcare, such as studying psychological and neurological disorders, and physiological conditions, and to evaluate the extent to which the human brain reacts to some stimuli. It then also goes on to introduce some concepts that are relayed in the entire report, such as Paradigms of emotions, Discrete Emotions Theory, and Multidimensional emotions theory. These topics are the basic foundations of the studies used in the report, so the introduction proficiently goes over them to allow the reader to fully understand the entirety of the report.
Methods and Materials
         This report styles its “Materials and Methods” section a bit differently. Because this report is a compilation of past research studies, instead of having a methods and materials section, this part of the lab report is broken down into many subsections that all talk about the different forms of data, data collection, and data selection. Each of these subsections is one of the previously mentioned forms of data, and the reason these were put here instead of a methods and materials section is that these are the methods of data collection and selection employed by the different lab reports that were used to compile this one. First, it talks about how the different lab reports got their information, through the use of subsections explaining the different methods such as Questionnaires, and the physical and physiological signals picked up by the AI such as speech, facial expressions, and Electroencephalogram (EEG) along others. The report then goes on to briefly discuss the automated emotion recognition systems and how they work, and what data they can read and find. Most of these automated emotion recognition systems get their information from the eyes, brain, heart, skin, and face, and this data is generated from our body’s reaction to stimuli. It then goes in-depth about the way of filtering and sorting that information to be as precise as possible.
 
 
Results
         The section that would be the results section isn’t there. Rather, this section is replaced with several subsections. The first subsection is the motivation and highlights of this review study, where it talks about the failures in previous studies where most failed to follow Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, failed to cover challenges in research, or only focused on one type of signal data. Normally, discussing the faults and shortcomings of previous studies comes in the introduction, but I believe that it was not included in the introduction because of 2 reasons. The first reason is that the section before this, what was supposed to be the materials and methods section, talks about how all the studies got their information. It would be illogical to explain the shortcomings of the previous studies, without explaining what each of the studies utilized and could have utilized, as the reader would not understand how those shortcomings like only using data from one signal type would have such an effect. The second reason is that this report is a compilation of many different reports, if you explain the shortcomings of the reports previously trying to compile this research information before explaining what is in this report, it may confuse with the reader as they can come to believe that those same problems that plagued the previous reports also affects this one.
         The next part of this section talks about the highlights and “results” of the study. The highlights of the study include the comprehensive use of many different datasets, how the review followed strict guidelines, how all the research they used was very recent, and how diverse the information was, using multi-modal emotion recognition (using different physiological and physical signals, paired with AI using machine learning and deep learning techniques). The study then goes on to talk about the “results”, which were the results of the individual studies and grouped with studies of the same nature, such as studies that used the same physiological signals as the basis of their study. These results include graphical data along with a summary of those pictures to help the reader understand what the actual results are.
Discussion
         Although the previous two sections deviated from what a normal lab report was to normally do, this section is much closer to what a discussion section in a standard lab report is than the previous two sections are to their standard counterparts. The discussion section is meant to put the data from the results into the context of the lab report and the reason for writing it. In this report, the discussion section uses the graphs and pictures provided with the data from the results and explains them. It groups up the studies based on the type of signal they were focused on and gives a detailed explanation of what can be learned from them. The report then goes on to give an overall discussion of automated emotion recognition systems, the main topic of this report, stating each signal type and its flaws and advantages. It then goes on to list all the different challenges faced when making the report, such as lack of data outside labs, the extensive testing of the available data making it hard for any new information to come, the lack of generalization within the methods, and signals of all the studies making it difficult to compare and get significant results. The report then goes on to give future recommendations and research directions by saying how automated emotion recognition should be used in the field to fully test and develop it, noting how it could be used in healthcare, environmental health, and assistance to humans in education, marketing, and business.

Conclusion
         The conclusion in this report is also fairly standard. A standard conclusion should summarize the main points of the report, the purpose of the research, and the findings. It also shouldn’t bring up any new information. This lab report’s conclusion does exactly that. It briefly goes over that this report was compiled due to the importance of emotion recognition in many different fields, and how deep learning models of automated emotion recognition are outclassing those of machine learning models. It also stresses the importance of transparency and exploitability to allow more people to utilize it and allow for more data to be gained for further research.

References
The reference section does its job as a reference section. It lists the citations for each of the lab reports and studies used in this lab report. There is also an appendix stating the deep numerical data for each of the lab reports.

This lab report gives us an insight into the research done on automated emotion recognition systems from the past 10 years. Due to this lab not being a typical lab, being a compilation of different lab reports and studies on automated emotion recognition systems, the lab report does not follow the standard format of a lab report outlined in Chapter 19 of “Technical Communication” by Mike Markel, but rather deviates from it slightly during the methods and materials and results section to compensate for the lack of the lab being a standard lab, adapting it to fit the style that a compilation of lab reports should have.

Citations:
Khare, S. K., Blanes-Vidal, V., Nadimi, E. S., & Acharya, U. R. (2024). Emotion recognition and Artificial Intelligence: A systematic review (2014–2023) and research recommendations. Information Fusion, 102, 102019. https://doi.org/10.1016/j.inffus.2023.102019

Markel, M., & Selber, S. A. (2021). Technical communication. Bedford/St. Martin’s, a Macmillan Education imprint.