# What is Wrong with Topic Modeling? (and How to Fix it Using Search-based Software Engineering)

arXiv: https://arxiv.org/abs/1608.08176 · HTML: https://mlsanity.com/paper/1608.08176/

## Key facts

- Cited by 14 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 Aug 2016.

## Built on by (most cited first, 14 of 14)

- [Are Neural Topic Models Broken?](https://mlsanity.com/paper/2210.16162/) (2022)
- [Comparison of Topic Modelling Approaches in the Banking Context](https://mlsanity.com/paper/2402.03176/) (2024)
- [No more Reviewer #2: Subverting Automatic Paper-Reviewer Assignment using Adversarial Learning](https://mlsanity.com/paper/2303.14443/) (2023)
- [Topic Modelling on Consumer Financial Protection Bureau Data: An Approach Using BERT Based Embeddings](https://mlsanity.com/paper/2205.07259/) (2022)
- [Determination of the Number of Topics Intrinsically: Is It Possible?](https://mlsanity.com/paper/2406.10402/) (2024)
- [HTMOT : Hierarchical Topic Modelling Over Time](https://mlsanity.com/paper/2112.03104/) (2021)
- [Active Learning for Identifying Disaster-Related Tweets: A Comparison with Keyword Filtering and Generic Fine-Tuning](https://mlsanity.com/paper/2408.09914/) (2024)
- [Are You Misinformed? A Study of Covid-Related Fake News in Bengali on Facebook](https://mlsanity.com/paper/2203.11669/) (2022)
- [When Less is More: On the Value of "Co-training" for Semi-Supervised Software Defect Predictors](https://mlsanity.com/paper/2211.05920/) (2022)
- [Assessing the Early Bird Heuristic (for Predicting Project Quality)](https://mlsanity.com/paper/2105.11082/) (2021)
- [Semiparametric Latent Topic Modeling on Consumer-Generated Corpora](https://mlsanity.com/paper/2107.10651/) (2021)
- [Using meaning instead of words to track topics](https://mlsanity.com/paper/2301.00565/) (2023)
- [SAP-sLDA: An Interpretable Interface for Exploring Unstructured Text](https://mlsanity.com/paper/2308.01420/) (2023)
- [Is Model Instability just Noise to be Tolerated or a Property that can be Managed?](https://mlsanity.com/paper/2607.10420/) (2026)

## Abstract (arXiv)

Context: Topic modeling finds human-readable structures in unstructured textual data. A widely used topic modeler is Latent Dirichlet allocation. When run on different datasets, LDA suffers from "order effects" i.e. different topics are generated if the order of training data is shuffled. Such order effects introduce a systematic error for any study. This error can relate to misleading results;specifically, inaccurate topic descriptions and a reduction in the efficacy of text mining classification results. Objective: To provide a method in which distributions generated by LDA are more stable and can be used for further analysis. Method: We use LDADE, a search-based software engineering tool that tunes LDA's parameters using DE (Differential Evolution). LDADE is evaluated on data from a programmer information exchange site (Stackoverflow), title and abstract text of thousands ofSoftware Engineering (SE) papers, and software defect reports from NASA. Results were collected across different implementations of LDA (Python+Scikit-Learn, Scala+Spark); across different platforms (Linux, Macintosh) and for different kinds of LDAs (VEM,or using Gibbs sampling). Results were scored via topic stability and text mining classification accuracy. Results: In all treatments: (i) standard LDA exhibits very large topic instability; (ii) LDADE's tunings dramatically reduce cluster instability; (iii) LDADE also leads to improved performances for supervised as well as unsupervised learning. Conclusion: Due to topic instability, using standard LDA with its "off-the-shelf" settings should now be depreciated. Also, in future, we should require SE papers that use LDA to test and (if needed) mitigate LDA topic instability. Finally, LDADE is a candidate technology for effectively and efficiently reducing that instability.