TY - GEN
T1 - LDPP at the FinNLP-2022 ERAI Task
T2 - 4th Workshop on Financial Technology and Natural Language Processing, FinNLP 2022
AU - Trust, Paul
AU - Minghim, Rosane
N1 - Publisher Copyright:
©2022 Association for Computational Linguistics.
PY - 2022
Y1 - 2022
N2 - Social media and online forums have made it easier for people to share their views and opinions on various topics in society. In this paper, we focus on posts discussing investment related topics. When it comes to investment, people can now easily share their opinions about online traded items and also provide rationales to support their arguments on social media. However, there are millions of posts to read with potential of having some posts from amateur investors or completely unrelated posts. Identifying the most important posts that could lead to higher maximal potential profit (MPP) and lower maximal loss for investment is not a trivial task. In this paper, propose to use determinantal point processes and variational autoencoders to identify high quality posts from the given rationales. Experimental results suggest that our method mines quality posts compared to random selection and also latent variable modeling improves improves the quality of selected posts.
AB - Social media and online forums have made it easier for people to share their views and opinions on various topics in society. In this paper, we focus on posts discussing investment related topics. When it comes to investment, people can now easily share their opinions about online traded items and also provide rationales to support their arguments on social media. However, there are millions of posts to read with potential of having some posts from amateur investors or completely unrelated posts. Identifying the most important posts that could lead to higher maximal potential profit (MPP) and lower maximal loss for investment is not a trivial task. In this paper, propose to use determinantal point processes and variational autoencoders to identify high quality posts from the given rationales. Experimental results suggest that our method mines quality posts compared to random selection and also latent variable modeling improves improves the quality of selected posts.
UR - https://www.scopus.com/pages/publications/85153178777
U2 - 10.18653/v1/2022.finnlp-1.18
DO - 10.18653/v1/2022.finnlp-1.18
M3 - Conference proceeding
AN - SCOPUS:85153178777
T3 - FinNLP 2022 - 4th Workshop on Financial Technology and Natural Language Processing, Proceedings of the Workshop
SP - 136
EP - 140
BT - FinNLP 2022 - 4th Workshop on Financial Technology and Natural Language Processing, Proceedings of the Workshop
PB - Association for Computational Linguistics (ACL)
Y2 - 8 December 0202
ER -