id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-189256-72eumkal	Santosh, Roshan	Detecting Emerging Symptoms of COVID-19 using Context-based Twitter Embeddings	2020-11-08		.txt	text/plain	2869	178	56	More generally, the method can be applied to finding context-specific words and texts (e.g. symptom mentions) in large imbalanced corpora (e.g. all tweets mentioning #COVID-19). We evaluate our graph-based approach on 2 different datasets, with each dataset having a different context -1) COVID-19 Symptom Detection; and 2) Adverse Drug Reaction Identification. With the 1 million COVID-19 tweet dataset, we use cough as the seed word, k = 0.3, maxDepth = 3 and n = 5 to test our approach. Evaluation Similar to COVID-19 symptom detection evaluation, we evaluate the model's performance for ADR detection, where a positive word represents an adverse reaction to a drug (Table 3) . In this study, we present an iterative learning approach to generate such a "master" list of COVID-19 symptoms, using the identification of words matching a specific symptom context.	./cache/cord-189256-72eumkal.txt	./txt/cord-189256-72eumkal.txt