In the dynamic realm of academic research, exceptional minds constantly push the boundaries of knowledge, unveiling breakthroughs that shape our understanding of the world. Barry Menglong Yao, a Ph.D. student at Virginia Tech under the guidance of Dr. Lifu Huang, stands as a testament to the spirit of relentless exploration and innovation.
With a rich academic background that encompasses both the University at Buffalo and Virginia Tech, Barry has garnered recognition for his exemplary research contributions.
Barry Menglong Yao received the distinguished Best Paper Award Honorable Mention from SIGIR 2023 and the prestigious Best MS Research Award from the Computer Science and Engineering department at the University at Buffalo.
Beyond the academic sphere, Barry finds solace and inspiration in the embrace of nature, a passion reflected in his love for hiking. Scaling towering peaks like Mt. Marcy and Mt. Siguniang Sanfeng, he finds parallels between the challenges faced in his academic journey and those encountered amidst the majestic landscapes.
In this exclusive interview, we unravel the layers of Barry Menglong Yao’s academic odyssey, delving into his best research paper experiences, and the insights he graciously shares with fellow researchers and aspiring academics. Join us as we explore the mind of an accomplished scholar, gaining valuable perspectives on academia, research, and the power of perseverance.
I have written an article on ” 08 Untold Secrets Revealed for Winning Best Paper Award”. This article will help you in writing a quality paper that will have a high probability of winning Best Paper Award.
A Conversation with Best Paper Award Honorable Mention: Barry Menglong Yao
Welcome, Barry, and thank you for being a part of this enlightening conversation!
Introduction and Inspiration:
1. Can you share a bit about yourself and your journey as a researcher? Any interesting hobbies or anecdotes you’d like to share?
I am Barry Menglong Yao, a Ph.D. student at Virginia Tech, advised by Dr. Lifu Huang. Previously, I was a master’s student at the University at Buffalo, advised by Dr. Changyou Chen. I was also a graduate research assistant in Dr. David Doermann’s Federated Learning project. I received the Best Paper Award Honorable Mention from SIGIR 2023 and the Best MS Research Award from the CSE department at the University at Buffalo.
Personally, I like hiking and reading. I climbed Mt. Marcy and Mt. Siguniang Sanfeng.
2. Congratulations on receiving the Best Paper Award in the 46th International ACM Research and Development in Information Retrieval Conference with Paper Titled: End-to-End Multimodal Fact-Checking and Explanation Generation: A Challenging Dataset and Models
Thanks. I am extremely honoured to receive the Best Paper Award Honorable Mention.
3. How did you feel when you first found out about the award, and what significance does it hold for you?
I was excited to hear our paper was selected for the awardee list. Walking toward the podium, I felt every piece of effort was worth it. This recognition is important because it reinforces my confidence in pursuing novel and high-quality research.
4. What inspired you to undertake this particular research project, and how did you come up with the research question?
Misinformation has been a growing public concern in society and has caused difficulty in finding reliable information online. Although some fact-checking works exist to fight misinformation, there is still a significant gap in applying automatic fact-checking in the real world. (1) While current studies predict a truthfulness label, it is also necessary to provide a textual statement to explain the prediction. These explanations are vital to justify how the conclusion is reached step by step based on external evidence and provide the public with a rationale to analyze the reasoning process and share it with others. (2) Some prior studies assume that a short piece of evidence text is already identified, based on which the models can directly predict the truthfulness of the target claim. However, this is not realistic in practice as the claim does not come with evidence, which should be retrieved from a knowledge base or the Internet. (3) Most previous studies only consider text while ignoring online articles’ multi-media nature (e.g., images).
To fill this gap, we propose an end-to-end multimodal fact-checking and explanation generation task, where the input is a claim and an extensive collection of web sources, including articles, images, videos, and tweets, and the goal is to assess the truthfulness of the claim by retrieving relevant evidence and predicting a truthfulness label (e.g., support, refute or not enough information), and to generate a statement to summarize and explain the reasoning and ruling process. To support research in this direction, we construct a benchmark dataset named Mocheg and experiment with several state-of-the-art neural architectures to serve as baseline performance.
Research Methodology and Challenges:
5. Could you provide an overview of the methodology you used in your study?
To establish the baseline performance on Mocheg, we design a framework for the End-to-End Multimodal Fact-checking and Explanation Generation task, which consists of three components for the corresponding sub-tasks. For Evidence Retrieval, an SBERT-based bi-encoder and a BERT-based cross-encoder are applied in a retrieve-then-rerank pipeline to retrieve the text evidence from the knowledge base, and a CLIP-based bi-encoder is applied to retrieve the image evidence from the knowledge base. For Claim Verification, we detect stances of text/image evidence towards the claim and fuse the stance of all evidence to predict the truthfulness. For Explanation Generation, we generate an explanation conditioned on the input claim, its truthfulness label, and text evidence. To guarantee the consistency between the gold truthfulness label and the generated explanation, we apply a pre-trained classifier to predict a truthfulness label based on the generated explanation to obtain the alignment reward and train our generation model with a policy gradient.
6. What were some of the key challenges you faced during the research process? How did you overcome them?
The hardest part may be to figure out a novel, exciting research question. In the beginning, you have to read many surveys and papers published in the last two or three years to be familiar with this general research topic and find some potential directions to further explore as future work. What are the limitations? How can we address them? These are questions we have to keep in mind when reading papers. After you get some ideas, you have to double-check whether someone has already explored your ideas, and you need to refine your initial idea with more details so that you can confirm whether the overall approach makes sense. At that point, your advisor and labmates will also step in. You must have thorough discussions with your advisor and labmates since they will be the first round of reviewers to verify your idea. After all of these are done, congratulations, you have opened a door to the research world.
Collaboration and Teamwork:
7. Many successful research projects involve collaboration. Could you tell us about your team and how collaboration played a role in the success of your project?
Our team is composed of two students and three professors. All these professors have provided valuable suggestions along the project from different views and significantly improved the paper writing and experiment designs. I still remember my shock when I faced my first draft, which was filled with their comments and modifications. My collaborator, Aditya, also helped build the initial version of the claim verification model and human performance experiments. He is also good at drawing figures and helped to create the initial figures for our model architecture, which my advisor and I further improved.
Key Findings and Impact:
8. What were the most significant findings or results of your research, and why do you think they were impactful?
This work proposes an end-to-end multimodal fact-checking and explanation generation task. To support research in this direction, we construct a benchmark dataset named Mocheg and design a framework for this task to serve as baseline performance. Our work paves the way to apply end-to-end automatic fact-checking in the real world, which is critically important since the sheer scale of misinformation has been overwhelming for human fact-checkers. The situation has become even more complicated with the emergence of generation models, like ChatGPT and Stable Diffusion since they could be intentionally misused to generate misinformation or wrongly spread misinformation due to the hallucination issue.
Paper Writing and Presentation:
9. How did you approach the paper-writing process, and do you have any tips for researchers on structuring and presenting their work effectively?
We have revised the draft for at least five rounds. The first round was pretty hard since it was my first time to write a research paper. I tried my best to formulate our work as one research story. Unfortunately, the story and the writing were awful initially, and my advisor, Dr. Lifu Huang, had to revise it paragraph by paragraph. At the same time, he left tons of comments on my draft for me to further improve it. I still remember that his comments are in red. When I looked at my first draft, I was facing the red fall foliage forest, which is full of red comments. The later revision is easier since more and more concerns have been addressed, and you begin to feel that it looks like good work now. Dr. Lifu Huang and other professors provided lots of comments for the following round of revision, and we finally got the current one.
Here I advise my readers to visit my blog post on “Writing an Effective Research Paper with 11 Major Sections“. This article will help you in writing research papers effectively with a high probability of acceptance.
10. Presenting research findings is as important as the paper itself. Could you share your strategies for making engaging and effective presentations at conferences or in journal presentations?
Firstly, try to remove content from your presentation. The presentation should interest others in your story instead of telling them technical details. If they are interested in your work, they can read your paper.
Secondly, practice as many times as possible. You should have at least one public rehearsal with your lab and advisor. Before that, record your presentation by yourself and check whether there are any issues. You can also practice with your friends.
Visit my article on 7-Step Method for Creating a Powerful Research Paper Presentation at Conferences. This article will help you in preparing powerful research paper presentation from your research paper.
Publication and Ethical Considerations:
11. Choosing the right conferences for presenting research can greatly influence its impact. How did you go about selecting the conferences where you submitted your work, and what factors did you consider?
Typically, our lab only submits papers to the top conferences, like the conference on this website: https://github.com/lixin4ever/Conference-Acceptance-Rate. Each conference focuses on its corresponding domains like NLP, CV, ML, information retrieval, etc. Based on the topic of our work, it may fit several conferences, and we choose the closest top conference.
Visit my article on “Avoiding Predatory Conferences and Journals: A Step by Step Guide for Researchers”. This article will help you in choosing the right conference for your research paper.
12. Many journals require authors to transfer copyright to the publisher. How did you navigate copyright issues, and did you consider open-access options for your paper?
We grant non-exclusive permission for ACM to include our paper in any and all forms, in SIGIR publication. We further grant permission for ACM to distribute or sell this submission as part of the SIGIR publication in electronic form, and as part of the ACM Digital Library, compilation media (CD, DVD, USB) or broadcast, cablecast, laserdisc, multimedia, or any other media format now or hereafter known.
Yes. Our paper is open-access. Through an agreement with ACM, Virginia Tech University’s libraries have paid for us to make our article Open Access at no cost to us.
Visit my articles on “Copyright Transfer and Granting Exclusive Rights: Key Differences for Research Authors” and “Open Access Journals: What do you Need to Know as a Researcher?” for further details on copyright and open-access publishing.
13. Ethical research is paramount. Can you share insights on how you ensured the integrity of your research, particularly in terms of plagiarism prevention and adherence to ethical guidelines?
If your work depends on any source, like some data crawled from some websites, you need to obtain their permission to publish the data for research purposes. You also need to check whether there are any biases in these data.
Visit my article on “The Consequences of Plagiarism: What You Need to Know?”. This article will help you in avoiding plagiarism in your article.
Quality Assurance:
14. The quality of writing in a research paper is crucial. What steps did you take to ensure the clarity and accuracy of your paper, including proofreading and revisions?
We have conducted at least five rounds of revision. All collaborators have provided valuable comments to improve the paper writing. I also applied Grammarly for proofreading.
Visit my article on “Importance of Proofreading in Research Paper Writing”. This article will help you understand the importance of proofreading and writing error-free articles.
Technology and Tools:
15. Were there any specific software, tools, or platforms that you found particularly helpful during your research phase? How did these tools contribute to the efficiency of your work?
Draw.io is applied to draw figures on our paper. Wandb is used to collect experiment results since it can show all experiment results in one table for us to organize experiments conveniently. Grammarly is a great tool for proofreading.
16. For the data analysis part of your research, what software or tools did you rely on? How did these tools assist in processing and analyzing your data effectively?
Wandb is applied to organize the experiment results and display the training details like training loss. For error analysis, we visualize all elements of error instances in an HTML format to quickly analyse the error cause.
17. Managing references is crucial in academic writing. Could you share your preferred reference management tool or method and how it simplified the citation process for your paper?
I am using Papers, which is similar to Mendeley. I tried both and think both are great for reference management. Since we read lots of papers along with our research, they can help us organize these papers hierarchically based on their topic and domain. It is also convenient that they will keep our comments on the paper. Finally, it is easy to export the references for all papers on one specific topic and add them to your paper draft.
18. How did you efficiently conduct your literature review? Were there any tools or strategies that helped you discover and organize relevant literature for your research?
As I mentioned before, I use Papers to organize relevant literature. Another tool I want to recommend is Google Scholar since you can add an alert there to follow the new paper on your selected topic. Paperwithcode is also an awesome platform to find some interesting papers on popular topics.
19. Effective data visualization can enhance the clarity of research findings. What tools or techniques did you employ for creating compelling data visualizations in your paper?
We mainly use the table with latex and figures drawn by the draw.io tool. During the research, we also visualize our data and results in HTML.
Visit my articles on “Maximizing the Impact of Your Research Paper with Graphs and Charts” and “Best Practices for Designing and Formatting Tables in Research Papers”. These articles will help you in applying effective visualization techniques for your research paper.
20. What software did you use for creating presentations of your research findings? Were there any features or tips that made your presentations more engaging?
I use Office 365. The key is how to formulate your work as one coherent and reasonable story instead of copying all content from your paper.
21. Managing a research project involves various tasks and timelines. How did you ensure efficient project management, and did you use any project management tools?
I use Workflowy to record the weekly process, meeting notes, and other content related to this project. I use Pomotodo to focus on the current task. Generally speaking, you work for 25 minutes and then have a break for 5 minutes. This kind of schedule helps me to keep energy throughout the day to be productive.
22. Keeping track of revisions and changes in your research and paper is essential. Did you use version control systems or methods to manage different versions of your work?
We use Overleaf for paper writing and GitHub for coding.
23. In the spirit of open science, did you employ any tools or practices to make your research more transparent, reproducible, or accessible to others in the academic community?
We release our dataset, code, and model checkpoints to make it more accessible.
24. Did you share your research data publicly? If so, which platforms or repositories did you use for data sharing?
Yes. It is hosted on our server with a public link to download. We chose to host it there since it is easier to improve the dataset in the future.
Reflection and Future Work:
25. Looking back, what personal lessons have you learned from this research journey? Any memorable moments or anecdotes you’d like to share?
Keep working! It will pay back.
26. As we conclude, what advice would you give to early-career researchers who aspire to achieve similar success in their academic careers?
For your first work, try to work with a hands-on mentor. If you are working with the assistant professor, you can ask whether they are hand-on or hand-off. If you are working with the (associate) professor, you can ask whether there will be a postdoc or senior Ph.D. student working with you since the (associate) professor is normally too busy to work closely with you, which is necessary for the first-time researcher.
27. Lastly, are there any individuals or organizations you’d like to acknowledge for their support, and can you give us a glimpse of your future research endeavours?
Thank my advisor, Dr. Lifu Huang, and co-advisors, Dr. Lichao Sun and Dr. Jin-Hee Cho, for their valuable suggestions and support. Thank my collaborator, Aditya, for his great help in model building, figure drawing, and paper writing. This research is based upon work supported by U.S. DARPA KMASS Program # HR001121S0034.
Dr. Vijay Rajpurohit : As our conversation with Barry Menglong Yao draws to a close, we extend our heartfelt appreciation for the valuable insights and experiences he has generously shared. His journey, marked by determination, achievements, and a passion for both academic rigour and the great outdoors, is an inspiring narrative for aspiring researchers and scholars.
Barry’s advice to persevere, embrace challenges, and foster a collaborative spirit serves as a guiding light for those navigating the academic landscape. We have gained a deeper understanding of the person behind the research accolades – an individual driven not only by academic excellence but also by a genuine love for exploration and the quest for knowledge.
We thank Barry Menglong Yao for his time, candour, and the opportunity to delve into his remarkable academic odyssey. We look forward to witnessing the continued impact of his work and the indelible mark he leaves on the world of research and beyond.
Code, Dataset, Check Points, Poster, Paper, and Presentation Videos Used for Research
Within the realms of this groundbreaking research and its captivating presentation, lies a treasure trove of resources. The heart and soul of Barry Yao’s exploration encompass:
- Code: Meticulously crafted, a testament to the intricate workings of research.
- Datasets: Carefully curated repositories of information, vital for analysis and findings.
- Checkpoints: Key milestones and progress markers along this academic journey.
- Poster: A visual narrative, encapsulating the essence of research, speaking volumes in imagery.
- Paper: The academic parchment etched with profound discoveries and scholarly insights.
- Presentation Video: An immersive experience, bringing our research to life and enhancing understanding.
- Slides: Well-organized, eye-catching and self-explanatory slides with intricate details of implementation.
To unravel the intricacies of the scholarly journey of Barry Yao, we invite you to delve into these invaluable materials, waiting to be discovered and embraced, forging a deeper connection with the essence of research work.
Summary
Barry Menglong Yao’s journey through academia is a vivid testament to the fusion of dedication, innovation, and a thirst for knowledge. From his formative years as a master’s student at the University at Buffalo to his current pursuit of a Ph.D. at Virginia Tech, Barry’s academic expedition has been guided by the wisdom of esteemed mentors and fueled by his own innate curiosity.
Receiving accolades such as the Best Paper Award Honorable Mention from SIGIR 2023 and the Best MS Research Award reflects not only his exceptional research acumen but also the profound impact of his work within the academic community. His engagement with projects like the Federated Learning initiative under Dr. David Doermann’s guidance showcases his collaborative spirit and commitment to cutting-edge advancements.
Beyond the academic sphere, Barry’s love for hiking and reading unveils a multifaceted individual whose experiences in nature and literature enrich his perspective and undoubtedly influence the creative thinking that fuels his research.
As Barry looks to the future, his vision extends beyond personal success to a desire to contribute meaningfully to his field and inspire the next generation of researchers. His advice to aspiring academics echoes the principles that have guided his own journey – persistence, a willingness to embrace challenges, and a passion for creating transformative solutions.
In closing, we extend our gratitude to Barry Menglong Yao for generously sharing his insights and experiences. His story serves as an inspiration, reminding us that with dedication, passion, and an inquisitive mind, the possibilities within the realm of research are truly boundless.
