NumClaim: A Dataset for Investor's Fine-grained Claim Detection

Introduction

Numerals provide important information in financial narratives. Our statistical result in the financial analysis reports shows that over 58.47% of sentences contain at least one numeral. Without the numerals, lots of fine-grained information in the analysis reports will be lost. This phenomenon evidences the importance of the numerals in the financial narrative. Based on our observation, investors always make a claim with an estimation. This estimation can be a cue for detecting the investor's fine-grained claim. Therefore, we propose an expert-annotated dataset, NumClaim, for probing argument mining in the financial narrative. Among 5,144 instances in the NumClaim dataset, 23.78% and 76.22% of instances containing numerals are annotated as ``In-claim'' and ``Out-of-claim'', respectively.

Format

The NumClaim dataset is consisted of "file", "text", "numeral", "offset", "category" and "in-claim" in json format.

Example

{

'file': 20181210_Jih-Sun-Securities-I_-----2385-TT---_1.pdf'

'text': '群光2019年動能推測以影像模組、高階鍵盤表現較佳,公司影像模組客戶包括G客戶與R客戶、Logitech、SportDV等,鍵盤輸入動能則來自於高階鍵盤包括平板外接、LED發光與電競鍵盤等出貨。群光1Q~3Q18一般鍵盤收入YoY-5%~-10%,高階鍵盤收入則超逾YoY+20%,主要係受惠於微軟Surface與電競需求增溫,而其電競客戶則包括Corsair、Logitech等,另外MSI、ASUS與中國品牌亦為供應商。群光2019年預期以電競鍵盤、LED發光鍵盤因市佔率提升而表現較佳。日盛預估群光2018年EPS5.03元,並預估2019年稅後EPS5.72元,群光2H18毛利率回溫,本益比僅約10倍,落於歷史低檔,維持買進評等。',

'numeral': '20',

'offset': 137,

'category': 'relative',

'in-claim': 0

}

Download

Since the 'text' is collected from Bloomberg Terminal, we can only share the full text with the Bloomberg Terminal users who have the right to access the same reports. Before you sign the agreement, please check whether you have the right to access the reports that we used. How to check the license in Bloomberg Terminal

Please write us an email with the agreement. Click here to download the agreement of NumClaim.

Email Address: cjchen@nlg.csie.ntu.edu.tw

How to Cite the Corpus

Please cite the following paper when referring to the NumClaim in academic publications and papers.

Chung-Chi Chen, Hen-Hsen Huang, and Hsin-Hsi Chen. 2020. NumClaim: Investor's Fine-grained Claim Detection. In Proceedings of the 29th ACM International Conference on Information and Knowledge Management (CIKM 2020), Virtual Event, Ireland.

License

NumClaim is licensed under the Creative Commons Attribution-Non-Commercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) license.