# Multi-channel Reverse Dictionary Model

**Authors:** Lei Zhang, Fanchao Qi, Zhiyuan Liu, Yasheng Wang, Qun Liu, Maosong Sun  
**arXiv:** [1912.08441](https://arxiv.org/abs/1912.08441)  
**Date:** 2019-12-18  
**Summary page:** https://www.summarizepaper.com/en/paper/1912.08441v1/

## Key points

- The "Multi-channel Reverse Dictionary Model" addresses limitations of existing reverse dictionary methods
- Struggles with highly variable input queries and low-frequency target words
- Uses a sentence encoder and multiple predictors to identify different characteristics of the target word from the input query
- Evaluated on English and Chinese datasets, achieving state-of-the-art performance
- Surpasses popular commercial reverse dictionary system on human-written description datasets
- Quantitative analyses and case study showcase effectiveness and robustness of the model
- Code and data available on GitHub at https://github.com/thunlp/MultiRD
- Accepted by AAAI Conference on Artificial Intelligence 2020, highlighting significance in advancing reverse dictionary methodologies

## Summary

The "Multi-channel Reverse Dictionary Model" proposed by Lei Zhang, Fanchao Qi, Zhiyuan Liu, Yasheng Wang, Qun Liu, and Maosong Sun addresses the limitations of existing reverse dictionary methods. These methods struggle with highly variable input queries and low-frequency target words. The multi-channel model aims to tackle both issues simultaneously by using a sentence encoder and multiple predictors that identify different characteristics of the target word from the input query. It has been evaluated on English and Chinese datasets containing dictionary definitions and human-written descriptions. The experimental results demonstrate that the model achieves state-of-the-art performance, surpassing even the most popular commercial reverse dictionary system on human-written description datasets. Additionally, quantitative analyses and a case study have been conducted to showcase the effectiveness and robustness of the model. The code and data for this work are available on GitHub at https://github.com/thunlp/MultiRD. This research has been accepted by the AAAI Conference on Artificial Intelligence 2020, highlighting its significance in advancing reverse dictionary methodologies.