Theme 1: Conversational interaction with data

Natural language interfaces (NLIs) have been the ``holy grail'' of human-computer interaction and information search. Early efforts in building NLIs, often grounded in grammar-based approaches, faced challenges due to the complexity of natural language understanding. Subsequent machine learning techniques led to a resurgence of interest, enabling systems such as semantic parsers, question-answering platforms, and search interfaces. These systems aimed to interpret user intent, resolve ambiguity, and support iterative refinement through conversational interactions. With the advent of LLMs, however, this vision is becoming increasingly viable; visualization NLIs are gaining popularity both within academic research and in mainstream commercial tools.

However, many open questions still persist regarding both the interpretation of language input and the interface and interaction design within these systems.

The first theme of this workshop focuses on improving people's ability to engage with data and visualizations through language-based interfaces.

Theme 2: Integrating text with charts

When integrating visualization with text, the textual component is often either overlooked or at least undervalued when exploring novel ways of communicating through data. Research in natural language generation (NLG) has made tremendous progress, with algorithms now being able to translate text, summarize articles, and describe imagery with high accuracy. As a result, a growing number of interfaces and tools are incorporating NLG techniques to help readers interpret visualizations and aid data-driven communication, ranging from creating captions, narratives, and stories to describing takeaways from a chart or dashboard. More recently, LLMs have opened new possibilities for applications ranging from dialogue systems, contextualized summary generation, and explanation. However, the use of NLG techniques in the context of information visualization is relatively nascent both in terms of utility and generalizability.

The second theme in this workshop will focus on ways to better understand the importance and role of text in information visualization.

Theme 3: Enhancing data semantics with NLP techniques

NLP techniques (e.g., sentiment analysis, entity classification, topic modeling) have traditionally been used to facilitate the analysis of textual data for scenarios such as investigative analysis, political speech/debate summarization, and social media content analysis, among others. However, NLP algorithms also present an opportunity for systems to understand the semantics of an input dataset (e.g., detecting geographic locations, people names, or specific types of date strings). This semantic enrichment can be used for richer data summaries, automatic defaults in visual analytics tools, and richer data exploration. While intriguing in theory, the application of semantics toward richer analytical experiences is still an underexplored idea. With the rise of LLMs, there is a new opportunity to infer fine-grained semantics from both structured and unstructured data, as well as from user queries and conversational context.

The third theme of the workshop will delve into text visualization systems and approaches to add semantics as part of the visual analysis process.