MsrGenomicsKbcomp is a dataset focused on relation extraction and knowledge completion in the genomics domain, released by Microsoft Research in 2016. It aims to construct a gene regulatory relation knowledge base using text co-occurrence information from PubMed abstracts and official pathway databases, supporting knowledge graph completion research.

The dataset contains 16,259 samples (12,160 in the training set, 2,784 in the test set, and 1,315 in the validation set), with a data size of approximately 342 KB. The data originates from the NCI PID Pathway Interaction Database and PubMed abstracts, processed and annotated using the Literome tool, covering gene relation types such as Positive_regulation, Negative_regulation, and Family, providing high-quality benchmark data for biomedical natural language processing tasks.

Dataset Composition

• train.txt: Knowledge base training set triples, used for model training.

• valid.txt: Knowledge base validation set triples, used for model validation.

• test.text: Knowledge base test set triples, used for model testing.

• text.txt: Gene relation text triples extracted from PubMed abstracts, processed through entity linking and Literome.

Citation```bibtex

@inproceedings{toutanova-etal-2016-compositional, title = "Compositional Learning of Embeddings for Relation Paths in Knowledge Base and Text", author = "Toutanova, Kristina and Lin, Victoria and Yih, Wen-tau and Poon, Hoifung and Quirk, Chris", booktitle = "Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", month = aug, year = "2016", address = "Berlin, Germany", publisher = "Association for Computational Linguistics", url = "https://www.aclweb.org/anthology/P16-1136", doi = "10.18653/v1/P16-1136", pages = "1434--1444", }