Extracting entities and relations from unstructured text has attracted increasing attention in recent
years but remains challenging, due to the intrinsic difficulty in identifying overlapping relations
with shared entities.
TPLinker formulates joint extraction as a token pair linking problem
Experiment results show that TPLinker performs significantly better on
overlapping and multiple relation extraction, and achieves state-of-the-art performance on two
public datasets.
To mitigate the issue
Extracting entities and relations from unstructured texts is an essential step in automatic knowledge base
construction.
Traditional pipelined approaches
However, due to the complete separation of entity detection and relation classification, these models ignore the interaction and correlation between
the two subtasks, being susceptible to cascading errors.
building joint models
利用:employ;exploit;integrate;adopt
减轻:alleviate;To break the bottleneck of manual labeling
These methods in-
troduce exposure bias, which may cause the
models overfit to the frequent label combina-
tion, thus limiting the generalization ability.