Joint Recognition and Linking of Fine-Grained Locations from Tweets

Abstract

Many users casually reveal their locations such as restaurants, landmarks, and shops in their tweets. Recognizing such fine-grained locations from tweets and then linking the location mentions to well-defined location profiles (e.g., with formal name, detailed address, and geo-coordinates etc.) offer a tremendous opportunity for many applications. Different from existing solutions which perform location recognition and linking as two sub-tasks sequentially in a pipeline setting, in this paper, we propose a novel joint framework to perform location recognition and location linking simultaneously in a joint search space. We formulate this end-to-end location linking problem as a structured prediction problem and propose a beamsearch based algorithm. Based on the concept of multi-view learning, we further enable the algorithm to learn from unlabeled data to alleviate the dearth of labeled data. Extensive experiments are conducted to recognize locations mentioned in tweets and link them to location profiles in Foursquare. Experimental results show that the proposed joint learning algorithm outperforms the state-of-theart solutions, and learning from unlabeled data improves both the recognition and linking accuracy.

Challenges of Sentiment Analysis for Dynamic Events

“…An application that predicts the election result must consider each state's influence in the election by means of the number of electoral votes for that state. Many tools and approaches have been developed for both fine-grained [8] and coarse-grained [17] location identification in tweets for different purposes such as disaster management and election monitoring. In the latter case, the geographic location of a tweet or the user location in the profile can be used to estimate the user's approximate location.…”

DLocRL: A Deep Learning Pipeline for Fine-Grained Location Recognition and Linking in Tweets

“…Pair-wise [15] and global collective methods [11,12,32] have been applied to explore the coherence. Zhang et al [34] and Ji et al [14] observed that the geographical coherence is effective in location disambiguation task. Also, instead of tweet-level coherence, Li et al [18] utilized user-level coherence to facilitate entity disambiguation.…”

“…More specifically, we compare our model with [16] and [14] for recognition, and [30] for linking, respectively. Note that the performance of [14] on linking subtask is not provided since it is a joint model. For the whole task, we choose JoRL L as the baseline since it is the stateof-the-art joint model in supervised learning manner.…”

“…On the other hand, existing solutions [10,14,16] on these two tasks require a large set of features manually designed for each task and domain, which demands task and domain expertise. For example, Li et al [16] carefully designed lexical, grammatical, geographical and BILOU schema features (totally 11 features) to extract finegrained locations from tweets.…”

To Do or Not To Do: Distill crowdsourced negative caveats to augment api documentation

“…First, DISCA currently uses a simple name-matching strategy to select candidate sentences. Entity linking approaches (Ji, Sun, Cong, & Han, 2016;Ye, Xing, Foo, Li, & Kapre, 2016;Moro, Raganato, & Navigli, 2014) based on machine learning could improve the performance of DISCA, because these approaches can better handle API-mention variations and thus provide more candidate sentences for selection.…”