Unofficial slides: From Zero to Hero: Human-In-The-Loop Entity Linking in Low Resource Domains (ACL 2020)

For weekly presentation.

Transcript

1. 1/26 1 Unofficial slides @izuna385
2. 2/26 Summary • To tackle with the low-resource domain where datasets are limited, • propose new recommendation scheme for creating annotation, • which is agnostic to novel domains.
3. 3/26 Entity Linking(EL) • Link mention to specific entity in Knowledge Base 3 Beam, Andrew L., et al. "Clinical Concept Embeddings Learned from Massive Sources of Medical Data." arXiv:1804.01486 (2018). entity Knowledge Base
4. 4/26 Procedure for EL 4 1. Prepare Mention/Context vector based on annotated data 2. Learn/prepare Entity (and its feature) representation 3. Candidate generation 4. Linking
5. 5/26 Procedure for EL 5 1. Prepare Mention/Context vector based on annotated data 2. Learn/prepare Entity (and its feature) representation 3. Candidate generation 4. Linking Require large amount of labeled data
6. 6/26 Defect of previous EL requiring lots of domain-specific annotations. (e.g. [Gillick et al., ‘19] : 9 million) • They can’t cope with new domains and KB for them. • Prior-based candidate generation works only in general domain. [Gillick et al., ‘19]
7. 7/26 Previous Human-In-The-Loop EL requires pre-trained sequence tagger and index for candidates. • This can’t solve cold-start problem, where no annotation exists. • Also, previous studies only links againsts Wikipedia. (If we link mentions to Wikipedia, prior can be utilized.)
8. 8/26 There proposal • Can interactive annotation, during which candidate recommendation system are trained, improve annotation process for human?
9. 9/26 There proposal • Can interactive annotation, during which candidate recommendation system are trained, improve annotation process for human? Human-In-The-Loop Approach (proposed) ▪ Main focus is candidate ranking step after candidates are generated. ▪ All entities in ref-KB are supposed to have title and description.
10. 10/26 Procedure
11. 11/26 Procedure Their forcus Levenshtein distance based Fuzzy search (Next slide)
12. 12/26 (Supplement) Examples of fuzzy search • [Murty et al., ACL ‘18] N character-gram features. ・TFIDF character Ngram(N=1~5) + L2 normalized + cossim. • In this From-Zero-to-Hero paper, they leveraged WordPiece tokenization of BERT. • Still searching other examples ...
13. 13/26 Candidate Ranking • Handcraft feature-based ranking with LightGBM / RankSVM / RankNet. • For enhancing interactiveness and avoid slow inference, they avoided DNN models and utilized Gradient boosted trees.
14. 14/26 Used Features for Candidate Ranking • See the original paper. • Note: Although they used cos-sim. between sentence and entity label, BERT-encoders were not the fine-tuned one.
15. 15/26 Dataset • Because mentions in WWO have the extreme variance in surface form, Avg. Amb is lower. • For WWO and 1641, fuzzy search are conducted, which results in the increase of Avg. Cand. : both stands for same name
16. 16/26 Experiments Can interactive annotation, during which candidate recommendation system are trained, improve annotation process for human? • For validating their research question (as the following.) • Evaluation 1: Performance of Recommender 　・V.S. Non-interactive ranking performance 2: Simulating User annotation 3: Real User’s annotation performance 　・Speed and needness for searching query
17. 17/26 1: Automatic suggestion performance • Result High Precision: If the gold is included in cands, that method can specify the gold with high performance. High Recall : Does that CG method can catch gold in candidates? If recall is high, gold may be in cands with high confidence.
18. 18/26 1: Automatic suggestion performance • Result High Precision: If the gold is included in cands, that method can specify the gold with high performance. High Recall : Does that CG method can catch gold in candidates? If recall is high, gold may be in cands with high confidence.
19. 19/26 1: Automatic suggestion performance • Result High Precision: If the gold is included in cands, that method can specify the gold with high performance. High Recall : Does that CG method can catch gold in candidates? If recall is high, gold may be in cands with high confidence. For noisy text, use Levenshtein.
20. 20/26 2: Candidate ranking performance Note: ▪ Evaluation is different for each dataset. ▪ “MFLE” denotes gold freq. in training dataset.
21. 21/26 2: Candidate ranking performance |C|: Average available candidates. t: training time ・AIDA is biased with training set. ・LightGBM and RansSVM are fast enough to re-train after each user’s annotation, they say.
22. 22/26 2: Candidate ranking -feature importance- • Jaccard with AIDA is due to the fact that wikidata has low length of entity desc. • Levenshtein ML for WWO and 1641 is because for creating Human-in-the-Loop situation, they create KB from mentions in documents. • Sentence Cos-sim. is useful over three datasets.
23. 23/26 3: Simulation • Only in annotating training datasets, because generalizing model is out-of-corpus. • Seed annotation --> add them to the training sets for ranker → User add new annotation → ...
24. 24/26 3: Simulation • Only in annotating training datasets, because generalizing model is out-of-corpus. • Seed annotation --> add them to the training sets for ranker → User add new annotation → ...
25. 25/26 4: Real User Study • Their annotation recomender • Leads to 35% faster annotation, 15% less search queries.
26. 26/26 Conclusion • Newly proposed annotation recommender for EL. • Some feature-based analysis for noisy-text EL. • Human-in-the-loop approach is effective for training candidate ranker and recommendation.