Good-Enough Example Extrapolation

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 595–607 November 7–11, 2021. c⃝2021 Association for Computational Linguistics 595 Text2Mol: Cross-Modal Molecule Retrieval with Natural Language Queries Carl Edwards, ChengXiang Zhai, Heng Ji University of Illinois Urbana-Champaign {cne2, czhai, hengji}@illinois.edu Abstract We propose a new task, Text2Mol, to re- trieve molecules using natural language de- scriptions as queries. Natural language and molecules encode information in very differ- ent ways, which leads to the exciting but chal- lenging problem of integrating these two very different modalities. Although some work has been done on text-based retrieval and structure- based retrieval, this new task requires integrat- ing molecules and natural language more di- rectly. Moreover, this can be viewed as an especially challenging cross-lingual retrieval problem by considering the molecules as a language with a very unique grammar. We construct a paired dataset of molecules and their corresponding text descriptions, which we use to learn an aligned common semantic embedding space for retrieval. We extend this to create a cross-modal attention-based model for explainability and reranking by interpret- ing the attentions as association rules. We also employ an ensemble approach to integrate our different architectures, which significantly im- proves results from 0.372 to 0.499 MRR. This new multimodal approach opens a new per- spective on solving problems in chemistry lit- erature understanding and molecular machine learning.1 1 Introduction Discovering new properties and applications of dif- ferent molecules is critical for accelerating discov- ery in medicine and science. Existing databases contain tens of millions of molecules; PubChem (Kim et al., 2016, 2019) alone has 110 million com- pounds. Many information retrieval (IR) tools for chemistry rely on queries based on natural language descriptions of the molecules and existing chem- ical reactions. Hundreds of millions of possible molecules cannot all possibly undergo laboratory 1The programs and data are publicly available at github.com/cnedwards/text2mol for research purposes. Water is an oxygen hydride consisting of an oxygen atom that is covalently bonded to two hydrogen atoms. 1. 2. 3. H2O H2O2 C2H6O Query Figure 1: Given a natural language description of water, we want to rank the corresponding molecule H2O first among all the possible molecules. experimentation and be given attention by experts in order to create a description. To address this issue, it is critical to retrieve molecules directly from natural language descriptions. This approach allows newly discovered molecules to be easily integrated into the proposed IR framework. Our framework also allows for semantic-level search be- tween natural language descriptions and molecules as well as for query expansion within traditional chemistry information retrieval systems. Over the past several years, chemists have begun to rely increasingly on computational techniques for cataloging molecules and predicting chemical reactions, products, and properties, such as yield, toxicity, and water solubility (Wu et al., 2018; Glavatskikh et al., 2019; Coley et al., 2017; Ah- neman et al., 2018; Fooshee et al., 2018). However, natural language and molecules are very different modalities of data, which makes integrating them together a …

Content truncated to 500 words