Template for Writing an Abstract

There are many ways to write a compelling abstract. However, if you are struggling with writing yours, here's a default template on how you might write one.

• sentence 1: background
• sentence 2: missing gap (e.g. "however, there's a problem blah")
• sentence 3: why is this bad
• sentence 4: "In this paper, we propose SystemX"
• sentence 5, 6 (and/or 7): technical depth on what SystemX is. That is, what is the secret sauce that makes SystemX possible?
• sentence 8: benefits of SystemX, aka the primary contribution of the paper (note that this should echo sentence 3 in that SystemX should have addressed the problem stated in sentence 3)
• sentence 9: "We evaluate SystemX..."
• sentence 10: "Our results indicate that (probelm stated in sentence 3 is no longer a problem)..."

Note that although this is specific to writing an abstract, the flow and format should be extensible to the Intro section. An abstract can be considered as a summary of the Intro, and the Intro as a summary of the rest of the paper.

Examples:

A system paper (Dylan's Paper, in submission to VIS 2019):

1. Deep learning models require the configuration of many layers and parameters in order to get good results.
2. However, there are currently few systematic guidelines for how to configure a successful model.
3. This means model builders often have to experiment with different configurations by manually programming different architectures (which is tedious and time consuming) or relying on purely automated approaches to generate and train the architectures (which is expensive).
4. In this paper, we present \syss, a visual analytics tool that allows a model builder to discover a deep learning model quickly via exploration and rapid experimentation of neural network architectures.
5. In \syss, the user explores the large and complex parameter space for neural network architectures using a combination of global inspection and local experimentation.
6. Through a visual overview of a set of models, the user identifies interesting clusters of architectures.
7. Based on their findings, the user can run ablation and variation experiments to identify the effects of adding, removing, or replacing layers in a given architecture and generate new models accordingly.
8. As a result, a model builder can build deep learning models quickly, efficiently, and without manual programming.
9. We evaluate \sys through user studies with four experts in deep learning.
10. The results indicate that using \sys users are able to gain new understanding of the model space, and are able to use such insights to discover novel and improved neural architectures.

A study paper (Marianne's Paper, in submission to VIS 2019):

1. Progressive visualization is fast becoming a technique in the visualization community to help users interact with large amounts of data.
2. With progressive visualization, users can examine intermediate results of complex or long running computations, without waiting for the computation to complete.
3. While this has shown to be beneficial to users, recent research has identified potential risks when using progressive visualization.
4. For example, users may misjudge the uncertainty in the intermediate results and draw the wrong conclusion or see patterns during the progression that are not present in the final results.
5. In this paper, we conduct a comprehensive set of studies to quantify the advantages and limitations of progressive visualization.
6. Based on a recent report by Micallef et al., our study examines four types of cognitive biases that can occur with progressive visualization: neglect of probability bias, confirmation bias, clustering illusion, and information bias.
7. The results of the study indicate that users' biases can impact their speed, accuracy, and confidence while analyzing data with progressive visualization.
8. However, the amount of impact varies with the uncertainty in the underlying progressive computation and the visualization of the results.
9. These findings confirm earlier reports of the drawbacks of progressive visualization and that continued research into mitigating the effects of cognitive biases is necessary.

A technique paper (Brian's paper, in preparation):

1. Modern time-series visual analysis tools make use of automated detection algorithms to recommend potentially anomalous views to the analyst when exploring large-scale time-series data.
2. However, these automated techniques are not perfect.
3. They can recommend views that users do not perceive as anomalous (false positives) or hide views that a user would deem abnormal (false negatives).
4. In this paper, we introduce a novel anomaly detection technique, called \sys, that recommends views in time-series data that closely resemble the users expectations.
5. Our technique leverages recent advances from the field of function learning in cognitive science research.
6. Function learning has been demonstrated to explain human predictions about values of a mathematical function.
7. Our technique extends on this finding by applying function learning to predicting human judgments about anomalies in time-series data.
8. (MISSING a sentence 8 (benefits of using \sys))
9. We evaluate our technique through a series of controlled experiments using both synthetic and real-world time-series data.
10. Our results indicate that the accuracy of our technique is consistent with the latest results from function learning and can effectively predict human-judgments of anomalies.