# Comprehensible Context-driven Text Game Playing

arXiv: https://arxiv.org/abs/1905.02265 · HTML: https://mlsanity.com/paper/1905.02265/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 6 May 2019.

## Built on by (most cited first, 10 of 10)

- [Introspective Tips: Large Language Model for In-Context Decision Making](https://mlsanity.com/paper/2305.11598/) (2023)
- [A Systematic Survey of Text Worlds as Embodied Natural Language Environments](https://mlsanity.com/paper/2107.04132/) (2021)
- [Towards Socially Intelligent Agents with Mental State Transition and Human Utility](https://mlsanity.com/paper/2103.07011/) (2021)
- [Training Value-Aligned Reinforcement Learning Agents Using a Normative Prior](https://mlsanity.com/paper/2104.09469/) (2021)
- [Learning to Follow Instructions in Text-Based Games](https://mlsanity.com/paper/2211.04591/) (2022)
- [Generalization in Text-based Games via Hierarchical Reinforcement Learning](https://mlsanity.com/paper/2109.09968/) (2021)
- [Perceiving the World: Question-guided Reinforcement Learning for Text-based Games](https://mlsanity.com/paper/2204.09597/) (2022)
- [Situated Language Learning via Interactive Narratives](https://mlsanity.com/paper/2103.09977/) (2021)
- [Affordance Extraction with an External Knowledge Database for Text-Based Simulated Environments](https://mlsanity.com/paper/2207.00265/) (2022)
- [An Analysis of Deep Reinforcement Learning Agents for Text-based Games](https://mlsanity.com/paper/2209.04105/) (2022)

## Abstract (arXiv)

In order to train a computer agent to play a text-based computer game, we must represent each hidden state of the game. A Long Short-Term Memory (LSTM) model running over observed texts is a common choice for state construction. However, a normal Deep Q-learning Network (DQN) for such an agent requires millions of steps of training or more to converge. As such, an LSTM-based DQN can take tens of days to finish the training process. Though we can use a Convolutional Neural Network (CNN) as a text-encoder to construct states much faster than the LSTM, doing so without an understanding of the syntactic context of the words being analyzed can slow convergence. In this paper, we use a fast CNN to encode position- and syntax-oriented structures extracted from observed texts as states. We additionally augment the reward signal in a universal and practical manner. Together, we show that our improvements can not only speed up the process by one order of magnitude but also learn a superior agent.