[Paper Review] Facetize: An Interactive Tool for Cleaning and Transforming Datasets for Facilitating Exploratory Search

Anna Kokolaki, Yannis Tzitzikas|arXiv (Cornell University)|Dec 27, 2018

Data Quality and ManagementDecision Sciences19 references3 citations

TL;DR

Facetize is an interactive, user-friendly editor that enables non-technical users to clean, transform, and structure static or dynamic datasets (e.g., CSV, SPARQL results) into faceted-search-ready formats through guided, visual operations. It supports key data preparation tasks like filtering, value replacement, hierarchy and interval creation, derived attribute generation, and geographic tagging—demonstrating high usability in user evaluations with 70% success rate in task completion.

ABSTRACT

There is a plethora of datasets in various formats which are usually stored in files, hosted in catalogs, or accessed through SPARQL endpoints. In most cases, these datasets cannot be straightforwardly explored by end users, for satisfying recall-oriented information needs. To fill this gap, in this paper we present the design and implementation of Facetize, an editor that allows users to transform (in an interactive manner) datasets, either static (i.e. stored in files), or dynamic (i.e. being the results of SPARQL queries), to datasets that can be directly explored effectively by themselves or other users. The latter (exploration) is achieved through the familiar interaction paradigm of Faceted Search (and Preference-enriched Faceted Search). Specifically in this paper we describe the requirements, we introduce the required set of transformations, and then we detail the functionality and the implementation of the editor Facetize that realizes these transformations. The supported operations cover a wide range of tasks (selection, visibility, deletions, edits, definition of hierarchies, intervals, derived attributes, and others) and Facetize enables the user to carry them out in a user-friendly and guided manner, without presupposing any technical background (regarding data representation or query languages). Finally we present the results of an evaluation with users. To the best of your knowledge, this is the first editor for this kind of tasks.

Motivation & Objective

• To address the challenge of making plain-text datasets (e.g., CSV, SPARQL results) directly explorable through faceted search, especially for non-technical users.
• To identify core requirements for transforming raw datasets into user-friendly, faceted-search-ready formats.
• To design and implement an interactive editor that supports data cleaning, transformation, and enrichment without requiring programming or technical data knowledge.
• To evaluate the usability and effectiveness of the tool in real-world data preparation scenarios.
• To enable both static and dynamic dataset processing, including direct SPARQL endpoint integration.

Proposed method

• The system provides a visual, guided interface for users to perform data transformations such as row filtering, value editing, and attribute deletion or hiding.
• It supports the creation of hierarchical groupings (e.g., location hierarchies) and numeric intervals (e.g., price ranges) to improve facet usability.
• Users can define derived attributes (e.g., a new 'Pets and Smoking' flag) based on existing values through a rule-based interface.
• Geographic attributes (latitude/longitude) can be explicitly marked for spatial visualization in downstream faceted search systems.
• The tool supports both static datasets (e.g., CSV files) and dynamic datasets retrieved via SPARQL endpoints, with project persistence for incremental editing.
• Transformation workflows are managed through a project-based system that preserves user-defined configurations across dataset refreshes.

Experimental results

Research questions

• RQ1 How can non-technical users effectively clean and structure raw datasets for use in faceted search systems without writing code or using query languages?
• RQ2 What set of data transformation operations is most effective and intuitive for preparing datasets for exploratory search?
• RQ3 How does the usability of an interactive, guided editor compare to traditional data wrangling tools in real-world data preparation tasks?
• RQ4 To what extent can a tool like Facetize support both static and dynamic datasets (e.g., SPARQL query results) in a unified, user-friendly interface?
• RQ5 What is the impact of guided, visual transformation workflows on task completion success and error rates in data preparation?

Key findings

• Seventy percent of users successfully completed the full data transformation task, producing a dataset that was correctly loadable into a faceted search system like Hippalus.
• Users who spent 10–15 minutes on the task had the highest success rate, indicating optimal time investment for effective data transformation.
• The most successful users made 1–3 errors (35% success rate), while those with no errors had only 20% success, suggesting that minor errors did not hinder performance.
• Forty percent of users rated Facetize as 'Very Useful' and 60% as 'Useful', indicating strong perceived value among participants.
• Only 10% of users quit the task, and the majority of failures were due to misusing the system rather than inherent flaws in the interface.
• The tool successfully supported complex transformations such as hierarchical grouping, interval creation, and derived attribute generation, even for users without technical backgrounds.