Data-to-Text Generation Improves Decision-Making Under Uncertainty

  title={Data-to-Text Generation Improves Decision-Making Under Uncertainty},
  author={Dimitra Gkatzia and Oliver Lemon and Verena Rieser},
  journal={IEEE Computational Intelligence Magazine},
Decision-making is often dependent on uncertain data, e.g. data associated with confidence scores or probabilities. This article presents a comparison of different information presentations for uncertain data and, for the first time, measures their effects on human decision-making, in the domain of weather forecast generation. We use a game-based setup to evaluate the different systems. We show that the use of Natural Language Generation (NLG) enhances decision-making under uncertainty… 

Figures and Tables from this paper

Controlling Contents in Data-to-Document Generation with Human-Designed Topic Labels

A model to generate various summaries and to control their contents by providing the explicit targets for a reference to the model as controllable factors is developed.

Development of data-to-text (D2T) on generic data using fuzzy sets

The experimental results show that the information generated is proven to represent the data provided and can be understood by the level of students even at the elementary school level, and the computation time is quite good.

Evaluating the text quality, human likeness and tailoring component of PASS: A Dutch data-to-text system for soccer

An evaluation of PASS, a data-to-text system that generates Dutch soccer reports from match statistics which are automatically tailored towards fans of one club or the other, found that compared to human-written texts, computer-generated texts were rated slightly lower on style-related text components and slightly higher in terms of the correctness of given information.

Automated Insights on Visualizations with Natural Language Generation

This research aims to reduce the cognitive effort in understanding tabular data by combining charts with ranked natural language generated (NLG) bullet point statements that summarize the top takeaways.

Comparing informativeness of an NLG chatbot vs graphical app in diet-information domain

An NLG chatbot that processes natural language queries and provides insights through a combination of charts and text is presented and it is found that the conversational context improved users’ understanding of dietary data in various tasks and that users considered the chatbot more useful and quick to use than traditional apps.

An Abstractive Summarization Technique with Variable Length Keywords as per Document Diversity

This paper has employed text summarization by TF-IDF with PageRank keywords, sentence score algorithm, and Word2Vec word embedding to provide a stable measure of similarity as compared to existing forms of text summarizing, and proposes a new form of abstractive summarization.

Building a Persuasive Virtual Dietitian

The MADiMan system is described, a system that supports users in managing their diets while admitting diet transgressions and the basic mechanisms related to reasoning, data interpretation and content selection for a numeric data-to-text NLG system are introduced.

Deep Learning for Text Style Transfer: A Survey

A systematic survey of the research on neuralText style transfer, spanning over 100 representative articles since the first neural text style transfer work in 2017, is presented.



Natural Language Generation enhances human decision-making with uncertain information

It is shown that the use of Natural Language Generation (NLG) improves decision-making under uncertainty, compared to state-of-the-art graphical-based representation methods.

A Game-Based Setup for Data Collection and Task-Based Evaluation of Uncertain Information Presentation

A game-based setup is demoed for exploring the effectiveness of different approaches (graphics vs NLG) to communicating uncertainty in rainfall and temperature predictions and to serve as a task-based evaluation setup for Natural Language Generation (NLG).

Choosing words in computer-generated weather forecasts

Content Selection in Data-to-Text Systems: A Survey

This survey initially introduces the field of data-to-text generation, describes the general data- to-text system architecture and then it reviews the state-of-the-art content selection methods.

Computing with words in decision making: foundations, trends and prospects

An historical perspective of CW in decision making is presented by examining the pioneer papers in the field along with its most recent applications and different linguistic computational models that have been applied to the decision making field are explored.

Unsupervised Concept-to-text Generation with Hypergraphs

This work presents a joint model that captures content selection and surface realization in an unsupervised domain-independent fashion and defines a probabilistic context-free grammar that globally describes the inherent structure of the input (a corpus of database records and text describing some of them).

When a graph is poorer than 100 words: A comparison of computerised natural language generation, human generated descriptions and graphical displays in neonatal intensive care

It is suggested that NLG might offer a viable automated approach to removing noise and artefacts in real, complex and dynamic data sets, thereby reducing visual complexity and mental workload, and enhancing decision-making particularly for inexperienced staff.

Vagueness as Cost Reduction: An Empirical Test

Results indicate that people respond more quickly and accurately to vague linguistic expressions than to crisp numerical expres- sions, but that this benefit also accrues to precise terms that avoid numbers.

Learning Effective Multimodal Dialogue Strategies from Wizard-of-Oz Data: Bootstrapping and Evaluation

This work uses Reinforcement Learning (RL) to learn multimodal dialogue strategies by interaction with a simulated environment which is “bootstrapped” from small amounts of Wizard-of-Oz data, and shows that RL significantly outperforms Supervised Learning when interacting in simulation as well as for interactions with real users.

Generating Instructions in Virtual Environments (GIVE):A Challenge and an Evaluation Testbed for NLG

NLG is too diverse for a single “competition”, and there are no mutually accepted evaluation metrics, so it is suggested that all the positive aspects, and only a few of the negative ones, can be achieved by putting forth a challenge to the community.