Text analysis atoms improve large data workflows with fewer model calls
Semantic Atoms for Scalable Text Analysis in Data Analytics in the Era of LLM-Based Agents
Computation and Language
Summary
Lots of useful information for understanding data is hidden in text fields, like customer complaints. The authors look at how to mix AI language tools and exact computer code to analyze these texts effectively. They introduce two basic tools called semantic atoms—tagging and categorization—that break down text analysis into manageable parts. Their approach, implemented in TADA-Agent, handles large text data efficiently and reuses results to reduce repeated AI calls. Tests show their method keeps quality similar to existing tools but is much more efficient and preserves analysis steps better.
What this means in practice
- •For data science teams: Process large text columns in data tables more efficiently by dividing semantic tasks for better reuse and fewer AI model calls.
- •For business intelligence developers: Integrate AI and programmatic querying to handle customer feedback text dynamically within analysis workflows for richer insights.
Authors
Jinxiang Xie, Zihao Li, Rui Ding, Bartuer Zhou, Shi Han, Dongmei Zhang
Abstract
Much of the information needed for data analysis is buried in text columns: an analyst may infer customer frustration from complaint text and use it to explain resolution time. We call workloads in which processing text columns is central to subsequent data analysis TADA (\emph{Text Analysis in Data Analytics}). In an agentic setting, the user's query dynamically specifies this processing, and answering it interleaves LLM-based semantic steps with exact SQL or Python operations. This mixed execution raises two design questions: which work should be neural and which symbolic, and which semantic outputs later steps can reuse. We address them for agent-planned TADA workflows with two complementary, query-driven semantic atoms. Each atom takes a sequence of text items and a natural-language query specifying the semantic task: \textbf{Tagging} returns one string per input item, while \textbf{Categorization} returns $M$ category objects over $N$ input items. Tagging results are materialized as a row-aligned column and Categorization results as a category table; SQL or Python then performs exact computation over this exposed state. We realize this interface in TADA-Agent, whose atoms run inside generated Python programs and internally handle inputs beyond a reliable context window. We evaluate TADA-Agent on TABench, a benchmark of 40 complete analysis requests over spreadsheet tables spanning workflow composition, input--output structure, and literature-grounded analytical scenarios. Against a semantic-operator baseline and relative to a held-out TADA reference run, TADA-Agent delivers comparable row-level semantic quality, has higher complete-artifact preservation estimates, and uses over an order of magnitude fewer semantic-model requests under the evaluated native execution paths.