In order to properly determine which of several possible meanings an acronym A in sentence s has, any system that aims to find the correct meaning for A must understand the context of s. This paper describes the techniques we use for that problem for the SDU@AAAI benchmark in which context was provided in the form of sentences in which acronym A is present and defined. As a capsule summary of our results, Support Vector Machines with Doc2Vec techniques achieves a higher Macro F1-Measure score than Cosine similarity with Classic Context Vector techniques. Although these techniques usually work better with documents (i.e., many sentences rather than the one sentence offered in this benchmark), they achieved scores of Macro F1-Measure 86-89%. While these results were 5.65% worse than the best in the benchmark experiment, the high speed of our approach (max 0.6 seconds on average per sentence on a virtual machine allocated with 4 CPU cores and 32GB of RAM in a shared server) and the possibility that our methods are complementary to those of other groups may lead to high performance hybrid systems.
|Original language||English (US)|
|Journal||CEUR Workshop Proceedings|
|State||Published - 2021|
|Event||2021 Workshop on Scientific Document Understanding, SDU 2021 - Virtual, Online|
Duration: Feb 9 2021 → …
ASJC Scopus subject areas
- Computer Science(all)