Zhenduo Wang

45 posts

Zhenduo Wang

Zhenduo Wang

@Zhenduo_W

Ph.D. in computer science. Conversational Search.

Atlanta, GA Katılım Ekim 2017
152 Takip Edilen83 Takipçiler
Sabitlenmiş Tweet
Zhenduo Wang
Zhenduo Wang@Zhenduo_W·
Life update: I defended my PhD in Computer Science at the University of Utah. I am excited to work as a Postdoctoral Fellow at the Georgia Institute of Technology with @alan_ritter.
English
3
1
61
4.9K
Chuan Meng
Chuan Meng@ChuanMg·
Happy to share that our tutorial “Conversational Search: From Fundamentals to Frontiers in the LLM Era”, with Fengran Mo, @maliannejadi and Jian-Yun Nie, got accepted at @SIGIRConf #SIGIR2025! Look forward to engaging discussions on LLM-based and agentic conversational search.
Chuan Meng tweet media
Amsterdam, The Netherlands 🇳🇱 English
2
2
31
1K
Zhenduo Wang
Zhenduo Wang@Zhenduo_W·
@EllenRiloff Today, those who stand on sentences with # of puns=1 shall fall. We could automatically disqualify them because they make us sick.
English
0
0
0
37
Ellen Riloff
Ellen Riloff@EllenRiloff·
I love this sign so much, both as an animal lover and a NLP geek -- it's both literally true and sarcastic.😁
Ellen Riloff tweet media
English
2
1
20
860
Ivan Sekulić
Ivan Sekulić@puding_p·
@Zhenduo_W Thank you for tuning in! I'm glad to hear that our findings align. I've seen your preprint and I'm looking forward to reading it in-depth :)
English
1
0
2
52
Zhenduo Wang
Zhenduo Wang@Zhenduo_W·
@puding_p We've reached many similar conclusions as yours in the talk. Finetuned models are usually better in every aspect. LLMs hallucinate user responses that would not be answered by real humans. That just invalidates the purpose of user simulators.
English
1
0
3
64
Zhenduo Wang
Zhenduo Wang@Zhenduo_W·
1. Why existing methods struggle to simulate users during search. 2. How the task resembles question answering but is distinctive. 3. Why LLMs are still inadequate for reliable user simulation. Our paper is now available: arxiv.org/abs/2304.07944 Code: github.com/zhenduow/UserS… 3/3
English
0
0
0
161
Zhenduo Wang
Zhenduo Wang@Zhenduo_W·
In our recent work "An In-depth Investigation of User Response Simulation for Conversational Search" accepted by #TheWebConf2024 @QingyaoAi @zhichaoxu_ir, we investigate the challenges of user simulation for conversational search and present timely analyses of: 2/3
English
1
0
2
854
Zhenduo Wang
Zhenduo Wang@Zhenduo_W·
Conversational search systems are mostly limited to being trained and evaluated with artificial and recorded search logs, which fails to address the open-ended nature of conversations in reality. A key reason for this is the lack of reliable user simulation. 1/3
Zhenduo Wang tweet media
English
1
1
7
610
Ellen Riloff
Ellen Riloff@EllenRiloff·
@Zhenduo_W @alan_ritter Congratulations, Zhenduo!!! Or I guess I should call you Dr. Wang now. 😀 And a second congratulations on that awesome post-doc position!! Super happy for you.
English
1
0
5
233
Zhenduo Wang
Zhenduo Wang@Zhenduo_W·
Life update: I defended my PhD in Computer Science at the University of Utah. I am excited to work as a Postdoctoral Fellow at the Georgia Institute of Technology with @alan_ritter.
English
3
1
61
4.9K
Zhenduo Wang
Zhenduo Wang@Zhenduo_W·
@QingyaoAi Thank you very much, Qingyao. I feel honored and grateful to have done my PhD with you!
English
0
0
0
50
Qingyao Ai
Qingyao Ai@QingyaoAi·
Huge congratulations to my student @Zhenduo_W on acing his thesis defense! Now officially Dr. Wang!👨‍🎓Zhenduo has done many excellent work on conversational search, and will continue his research career as a postdoc. Wishing you continued success in all your future!🎉📚
Qingyao Ai tweet media
English
3
0
27
1.3K