Title: Statement of your core result or finding.

Try to make your title an assertive statement, such as:

• “Changes in cytoplasmic volume are sufficient to drive spindle scaling.”

and not

• “High-performance silicon photoanodes passivated with ultrathin nickel films for water oxidation”

Rule of thumb: if your title would look weird with a period at the end, it is probably a poor title.

Don’t do this.

Abstract

Try to tell a story here, no matter what your field. You are writing for human beings, not computers. What’s the area, what’s the problem you are trying to understand. How? What have you found?

(You are summarizing your core results, not cramming them into this tiny space).

target: 84-151 words

current: 43

Press w to get the word count.

Introduction - “The Setup”

[In field X, we still don’t understand Y & Z.]

Write a summary of the question(s) you are trying to answer.
What is the state of the world before your research came along?
Also, answer the harsh but important question: Who cares?

In writing this, you can start general, but make sure clearly define the “before” state of the world’s knowledge for the specific area this paper is addressing.

Intro - Assertive Statement 1

Here you can expand on your introduction. To guide your writing, title this card with assertive statements:
Instead of “Problem Description”, be direct: “The problem is that X doesn’t do Y.”

Applications and importance of AI in education

Adaptive Learning

Serious Games

Bellotti et al.’s model

• RL, adaptive
• serious games

Adopting an adaptive approach toward education

Adopting a gamified approach toward education

Motivation for playing games

Discussion

Results are objective, but science isn’t about listing data, it’s about extracting meaning from what we observe.

What do your results tell you about the core problem you were investigating?

Conclusion

Bring it back to the big picture. How do your results fit into the current body of knowledge?

Most importantly, how can these results help you ask better questions?

Conclusion (further detail)

Expand on your conclusion summary, and add more details to it.

Conclusion

Final text for conclusion goes here

in as many

cards as you like.

References

We don’t have bibliography support yet, but we do have “named links” so you can refer to specific links by name rather than retyping it each time.

“Black holes are cool.” [1], and DNA is cool too [2]. But black holes are still cool, though not “absolute zero” cool [1].

Annotated Bibliography

#review
[1]J. Beck, M. Stern, and E. Haugsjaa, “Applications of AI in Education,” Crossroads, vol. 3, no. 1, pp. 11–15, Sep. 1996.

Summary:

• topic: Intelligent Tutoring Systems (ITS): student model, pedagogical module, domain knowledge, expert model, communication model
• research question: review evaluation of existing ITS systems (as of 1996)
• conclusion: ITS have been shown to be useful for students’ learning. However, more research needs to be done to understand each of its modules. They suggest that future work in this field includes collaborative learning.

Evaluation:
(background/methods/results/conclusion)

• This article demonstrates that the concept of using AI in education was already researched to a rather sophisticated extent in 1996. For a long time, researchers have seen promise in this area.

Reflection (about how this study fits into my research):

• My research will focus on the student model. I would like to look further into collaborative learning, which can ease the requirements of the ITS system by introducing peers who can help the student learn.

#importance
[1]J. Kay, “AI and Education: Grand Challenges,” IEEE Intelligent Systems, vol. 27, no. 5, pp. 66–69, Sep. 2012.

Summary:

• topic: AI in Education (AIED), which is very similar to ITS
• research question: importance of AIED research via AIED topic mentions in Grand Challenges
• conclusion: “AIED has already achieved much.” but “Much work lies ahead in tackling the Grand Challenges.”

Evaluation:
(background/methods/results/conclusion)

• This study has established the importance of AIED research not only through its potential applications but also through real mentions of AIED topics in various Grand Challenges.

Reflection (about how this study fits into my research):

• This study can help me justify my research topic in AI and education as a topic that is relevant to today’s problems.
• The diagram of key elements in AIED depicts how various AI research topics fit into the AIED system. This understanding will help me find a topic that fits into the AIED ecosystem.

#learning_by_teaching
[1]F. Tanaka and S. Matsuzoe, “Learning Verbs by Teaching a Care-receiving Robot by Children: An Experimental Report,” in Proceedings of the Seventh Annual ACM/IEEE International Conference on Human-Robot Interaction, New York, NY, USA, 2012, pp. 253–254.

Summary:

• topic: learning by teaching a Care-Receiving Robot (CRR)
• research question: “the use of care-receiving robot (CRR) for the purpose of supporting childhood education”
• method: verb-learning game using cards
  • the children (subjects) are tasked to learn 4 verbs
  • for 2/4 verbs: the children are tasked to teach the verb to the CRR
• conclusion: the most important idea is “to accelerate children’s natural motivation to caretaking by introducing the CRR”
  • after learning the verbs (with or without CRR), the children were tested on these verbs: learning with CRR had a higher percentage of correct answers

Evaluation:
(background/methods/results/conclusion)

• In both cases of “with/without CRR”, the experimenter would demonstrate the current verb to the child. In the case of “without CRR”, the child would immediately go on to learning the next verb after the demonstration. However, in the case of “with CRR”, the child would continue to have further exposure to the current verb by teaching this verb to the CRR. Therefore, the longer exposure, rather than the CRR, itself, could account for the higher percentage of correct answers in the case of “with CRR”.

Reflection (about how this study fits into my research):

• The study notes that “[t]his experiment was not designed to compare human teachers with the CRR, and, thus, the results should not be interpreted as showing the superiority of the CRR to human teaching.” With the current state of AI research, I do not believe that it is feasible (or necessary) to replace teachers. Instead, technologies like the CRR could be a useful supplement to the teacher-student learning framework, because it can provide one-on-one attention to the student when the teacher is not present. Therefore, my research will focus on AI that can facilitate this kind of supplement.
• The CRR used in this study was teleoperated and did not have any learning capabilities of its own. My research will take this a step further by focusing on learning algorithms in AI, such as reinforcement learning.
• Despite the limitations of this study’s methods in showing the effectiveness of the CRR, I am interested in the concept of learning by teaching. On a high level, I am interested in researching the effectiveness of having a student teach AI that is modeled after themselves.

#reinforcement_learning #game
[1]F. Bellotti, R. Berta, A. D. Gloria, and L. Primavera, “Adaptive Experience Engine for Serious Games,” IEEE Transactions on Computational Intelligence and AI in Games, vol. 1, no. 4, pp. 264–280, Dec. 2009.

Summary:

• topic: Sand Box Serious Games (SBSG) and Adaptive Experience Engine (EE)
• research question: proposing an SBSG system that is divided into the authoring stage and the runtime execution (incl. EE)
• method: proposing model architecture, its various parameters, its cost functions, and implementations of its EE (genetic computation and reinforcement learning)
• conclusion: “SGs represent an opportunity for enhancing education, also in a lifelong learning perspective. However, design paradigms are needed in order to spur effective player’s cognitive processing and allow cost-effective development.”
  • Therefore, this study has proposed an SBSG model that incorporates task authoring instead of a complete narrative authoring. The player can then build his/her own narrative by interacting with the game environment and its tasks.
  • Potential benefits: SG design is grounded in pedagogical principles; authors can focus on content/strategy rather than (hard-coded) narrative; it is easy to author new tasks to be added to the model without needing any code changes (content agnostic); the tasks can be reused in different games; the model “is able to manage a database of … tasks and to consider a variety of user needs and teacher requirements” (via a domain agnostic and reusable EE); the model “scales with the number of tasks and users” (via a tree model of tasks)

Evaluation:
(background/methods/results/conclusion)

• The concept of sand box games is not novel, but this study has developed a novel approach to sand box games that facilitates a player’s exploration of some game environment that actually represents an educational knowledge space. In particular, this approach integrates human experts, who supply the units of the knowledge space, and personalizes each player’s exploration of this knowledge space by taking into account each player’s profile.
• This study is only a model proposal; it does not test this game model on students.

Reflection (about how this study fits into my research):

• I am researching how reinforcement learning can be applied to education technology, and this study is one relevant application: reinforcement learning is used to implement the EE, which decides the task sequence that facilitates the player’s exploration of the knowledge space.
• I cannot find any source code for Bellotti et al.’s proposed model, but their proposal is detailed enough for me to be able to reimplement and adjust their model for my purposes. I am especially interested in finding some educational application for their reinforcement-learning-based EE.
• My paper will likely consist of a model/product proposal, so Bellotti et al.’s paper is a useful reference for writing my paper.

Thoughts:

• Interactive storytelling architecture for training (ISAT) has an intelligent director agent that is responsible for personalizing the experience of the trainee/player (p266). What if this agent took on a role more like the CRR instead of a director role?
  • Create frontend with https://twinery.org and hack it to include python backend with AI? Python hacking via https://github.com/tweecode/twine
  • https://medium.com/columbia-dsl/30-immersive-storytelling-platforms-apps-resources-tools-e428309574be
• game dev using godot (open source, with python support): https://github.com/touilleMan/godot-python
• use EE algorithm for task scheduling/management?
  • so the student would take the role of the task author and game author as well as the player
• Different types of successful games
  • narrative (nonadaptive): Monument Valley
  • sand box (personalized experience from exploring and experimenting with the game environment): Minecraft
  • #TODO

Existing Tools

• My Learning Assistant (MyLA): inquiry management tool
• Delta3D: open source GE
• ALIGN: “intends to maximize reuse in developing adaptive SGs by focusing on provision of user motivational and hinting support and meta-cognitive feedback”
• RETAIN: “SG design paradigm aimed at applying instructional strategies concurrent to game development”
• Crystal Island: “environment that supports an inquiry- based approach, where the story could be described as a con- tainer of elements to be taken into consideration in order for the player to solve problems in the domain of biology”
• Thespian: “interactive drama system supporting goal- driven decision-theoretic multiple agents that are responsive to user’s interaction while maintaining consistency with their roles in the story”

#TODO
[1]K. E. Merrick and M. L. Maher, “Motivated Reinforcement Learning for Adaptive Characters in Open-ended Simulation Games,” in Proceedings of the International Conference on Advances in Computer Entertainment Technology, New York, NY, USA, 2007, pp. 127–134.

Summary:

• topic: non-player characters (NPC) ins simulation games, motivated reinforcement learning (MRL)
• research question: proposal for designing adaptive NPCs
  • “design of non-player characters which can respond autonomously to unpredictable, open-ended changes to their environment”
  • “representing unpredictable, evolving worlds for character reasoning” via MRL using context-free grammars (CFG)
• method: Merrik et al. demonstrated their proposal through a sheep herding game in Second Life, where they applied their algorithms to sheep NPCs
• conclusion: this study proposes “a model for adaptive non-player characters in open-ended virtual worlds”

Evaluation:
(background/methods/results/conclusion)

Reflection (about how this study fits into my research):

#TODO
[1]M. Sanders and A. George, “Viewing the changing world of educational technology from a different perspective: Present realities, past lessons, and future possibilities,” Educ Inf Technol, vol. 22, no. 6, pp. 2915–2933, Nov. 2017.

Summary:

• topic:
• research question:
• method:
• conclusion:

Evaluation:
(background/methods/results/conclusion)

Reflection (about how this study fits into my research):

#TODO
[1]M. El Fouki, N. Aknin, and K. E. El. Kadiri, “Intelligent Adapted e-Learning System Based on Deep Reinforcement Learning,” in Proceedings of the 2Nd International Conference on Computing and Wireless Communication Systems, New York, NY, USA, 2017, p. 85:1–85:6.

#TODO
[1]T. Mandel, Y.-E. Liu, S. Levine, E. Brunskill, and Z. Popovic, “Offline Policy Evaluation Across Representations with Applications to Educational Games,” in Proceedings of the 2014 International Conference on Autonomous Agents and Multi-agent Systems, Richland, SC, 2014, pp. 1077–1084.

#TODO
[1]C. Tekin, K. Moon, and M. van der Schaar, “Staged Multi-armed Bandits,” arXiv:1508.00641 [cs, stat], Aug. 2015.

#motivation #game
[1]N. Yee, “Motivations for Play in Online Games,” CyberPsychology & Behavior, vol. 9, no. 6, pp. 772–775, Dec. 2006.

Summary:

• topic: different motivations for playing games
  • specifically, Massively-Multiplayer Online Role-Playing Games (MMORPGs)
• research question: Yee wants to take a “factor analytic approach to creating an empirically grounded player motivation model”
  • (factor analysis just refers to PCA)
• method: 3000 players were given a survey asking about their motivations for playing games
  • these questions were implemented as a 5 point rating system, e.g. “how important is…?” —> rate
  • PCA revealed 3 main components for motivation: achievement, social and immersion
• conclusion: different people choose to play games for very different reasons

Evaluation:
(background/methods/results/conclusion)

• this paper is not a theory (e.g. Bartle’s model of player types is a theory) but is a big empirical study that gives us real insight into what motivates players

Reflection (about how this study fits into my research):

• Bellotti et al.’s model for serious games has some features of the achievement and immersion components. Thus, gamifying education in such a way could motivate people to learn.

Nonacademic Sources

#game
DMLResearchHub, Games and Education Scholar James Paul Gee on Video Games, Learning, and Literacy.

https://www.youtube.com/watch?v=LNfPdaKYOPI

Notes

• “assessment and testing is what drives our current school system”
  • so our school teaches us the way that it does to accomodate these tests, not necessarily the learning!
• why are there tests for algebra but no tests for Halo?: “you actually trust the design and learning of Halo better than you trust the design and learning of that algebra class”

How to write an academic paper

How to write an annotated bibliography - WRITING 101

1. Summarize: What topic is covered? What is the research question (i.e., the main argument)? If someone asked what the article is about, what would you say? Very briefly, how was the study conducted? What are the authors’ major conclusions based on the results?
2. Analyze: Now it’s time to evaluate the work. Consider the following questions and choose the top one (or two) issues to include in your annotation.
   • Background: How does this study fill a gap in knowledge? What makes this study unique?
   • Methodology: Are there aspects of the study methodology that seem questionable given the objectives of the research? Are there any confounding variables that the author has no considered?
   • Results and Conclusions: Do the results adequately support the stated claims? Are the claims overgeneralized in light of the study details? Can you think of other reasonable interpretations of the results? What conclusions do you think can be reasonably drawn given the data?
3. Reflect: Once you’ve summarized and evaluated a source, ask yourself how it fits into your research proposal. Is this source helpful to you? How does it help you shape your research question? How could you use this source in your proposal? Has it changed how you think about your topic?

Template

Summary:

• topic:
• research question:
• method:
• conclusion:

Evaluation:
(background/methods/results/conclusion)

Reflection (about how this study fits into my research):

How to write a introduction - WRITING 101

This section explains the purpose of the study while helping the reader understand what is currently known about the topic. It also details the hypotheses within the context of the background literature. A successful introduction will:

• Explain the problem that your proposal will investigate. What are the gaps in knowledge or understanding?
• State your thesis (i.e., purpose statement) forcefully and directly. What do you expect to gain from the study?
• Define all key terms and concepts while avoiding jargon.
• Establish context by giving sufficient background information (This is the “literature review.”).
  • How does your question connect to and extend previous research on your topic?
  • Demonstrate that you have a good knowledge of the literature related to your topic.
• Provide a 1-2 sentence overview or blueprint of how you will investigate the question (to be
  expanded upon in full detail in the Methods section). What variables will you measure and/or
  manipulate?
• State a clear, testable hypothesis. Be as specific as possible about the relationship between the
  different variables in the study.
  • Not a hypothesis: I hypothesize that there will be a significant relationship between
    meditation and anxiety levels.
  • Hypothesis: I hypothesize that, over a three month period, anxiety levels will
    significantly decrease in the group receiving daily meditation training, but not in the group receiving daily cognitive training with Lumosity.

How to use this template

The idea here is to start at the far left, and clarify what the core of what you want to say is first, and then expand on it by moving to the right, one column at a time.

After a couple of “passes” of expanding, you will end up with your complete, and well structured paper on column 5, which you can export separately.

Here’s a (somewhat dated) video which might help.