Real World Conversational Entity Linking Requires More Than Zeroshots
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866929482679451648 |
|---|---|
| author | Hoveyda, Mohanna de Vries, Arjen P. de Rijke, Maarten Hasibi, Faegheh |
| author_facet | Hoveyda, Mohanna de Vries, Arjen P. de Rijke, Maarten Hasibi, Faegheh |
| contents | Entity linking (EL) in conversations faces notable challenges in practical applications, primarily due to the scarcity of entity-annotated conversational datasets and sparse knowledge bases (KB) containing domain-specific, long-tail entities. We designed targeted evaluation scenarios to measure the efficacy of EL models under resource constraints. Our evaluation employs two KBs: Fandom, exemplifying real-world EL complexities, and the widely used Wikipedia. First, we assess EL models' ability to generalize to a new unfamiliar KB using Fandom and a novel zero-shot conversational entity linking dataset that we curated based on Reddit discussions on Fandom entities. We then evaluate the adaptability of EL models to conversational settings without prior training. Our results indicate that current zero-shot EL models falter when introduced to new, domain-specific KBs without prior training, significantly dropping in performance. Our findings reveal that previous evaluation approaches fall short of capturing real-world complexities for zero-shot EL, highlighting the necessity for new approaches to design and assess conversational EL models to adapt to limited resources. The evaluation setup and the dataset proposed in this research are made publicly available. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_01152 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Real World Conversational Entity Linking Requires More Than Zeroshots Hoveyda, Mohanna de Vries, Arjen P. de Rijke, Maarten Hasibi, Faegheh Computation and Language Information Retrieval Entity linking (EL) in conversations faces notable challenges in practical applications, primarily due to the scarcity of entity-annotated conversational datasets and sparse knowledge bases (KB) containing domain-specific, long-tail entities. We designed targeted evaluation scenarios to measure the efficacy of EL models under resource constraints. Our evaluation employs two KBs: Fandom, exemplifying real-world EL complexities, and the widely used Wikipedia. First, we assess EL models' ability to generalize to a new unfamiliar KB using Fandom and a novel zero-shot conversational entity linking dataset that we curated based on Reddit discussions on Fandom entities. We then evaluate the adaptability of EL models to conversational settings without prior training. Our results indicate that current zero-shot EL models falter when introduced to new, domain-specific KBs without prior training, significantly dropping in performance. Our findings reveal that previous evaluation approaches fall short of capturing real-world complexities for zero-shot EL, highlighting the necessity for new approaches to design and assess conversational EL models to adapt to limited resources. The evaluation setup and the dataset proposed in this research are made publicly available. |
| title | Real World Conversational Entity Linking Requires More Than Zeroshots |
| topic | Computation and Language Information Retrieval |
| url | https://arxiv.org/abs/2409.01152 |