Understanding and Improving Data Repurposing

B. Greenwood; C. Cooper; J. Parsons; R. Lukyanenko

REVIEW 3 major objections 2 minor 19 references

Understanding and Improving Data Repurposing

T0 review · 3 major / 2 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper defines data repurposing as using data after adding attributes or entity types to the original schema, distinguishes it from data reuse, and develops a framework of concepts and activities for managing repurposing.

desk verdict A useful synthesis with a load-bearing boundary problem: the reuse/repurposing distinction collapses when checked against the paper's own examples. read the letter →

arxiv 2506.09073 v1 pith:5EBTRA4O submitted 2025-06-09 cs.CY cs.HC

J. Parsons , R. Lukyanenko , B. Greenwood , C. Cooper This is my paper

classification cs.CYcs.HC

keywords datarepurposingreuseschemaaugmentationuse-agnosticqualitymanagementdata-taskalignmentelasticitytransparency

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This commentary establishes that data repurposing is a distinct practice: using existing data for new tasks after making one or more additions to the schema that describes the original data. It contrasts repurposing with original data use and with data reuse, where new questions can be answered without changing the schema. The paper then proposes a framework of concepts and activities for adapting existing data to unanticipated tasks, including feasibility assessment, task reconceptualization, acquisition, exposition, and transformation. If the framework holds, data management must expand beyond designing data for predetermined uses to actively supporting data that can be stretched to future purposes.

What carries the argument

The schema is the load-bearing object: it is the explicit or implicit conceptual blueprint that converts a task into data abstractions such as entity types, attributes, and cardinalities. Whether a new task can be served by the existing schema determines the reuse–repurposing boundary. The framework's supporting concepts are the use-agnostic properties accessibility, transparency, and elasticity, plus the activities feasibility assessment, task reconceptualization, acquisition, exposition, and transformation. Elasticity is the capacity of data to conform to a different schema, tied to conceptual data independence, and it is shaped by granularity, storage design, governance, and psychological anchoring.

What would settle it

A documented case where data are successfully adapted to a new task without any addition of attributes or entity types—for example, re-aggregating administrative records to a different granularity, or reinterpreting existing fields under a new definition—yet practitioners and outcome measures treat the activity as repurposing rather than reuse would break the proposed demarcation.

Watch

Extended reading notes

Core claim

The paper's central claim is that data repurposing occurs only when the new task requires augmenting the original data schema with additional entities, attributes, or relationships; data reuse, by contrast, covers new questions that the existing schema can already support. On this basis the paper develops a Data Repurposing Framework with three use-agnostic data properties—accessibility, transparency, and elasticity—and five repurposing activities: feasibility assessment, task reconceptualization, acquisition, exposition, and transformation. It argues that elasticity extends the classic notion of data independence to conceptual independence, meaning data can be decoupled from its original schema and aligned with a new one. The framework is illustrated with a task-driven citizen-science case and a data-driven healthcare case, showing how it can expose strengths, weaknesses, and hidden costs of repurposing.

Load-bearing premise

The framework assumes that the line between data reuse and data repurposing lies exactly at whether the existing schema suffices for the new task, with no empirical evidence that this boundary matches how practitioners distinguish the two.

Editorial extensions

If this is right

Data management must broaden from original task design to include repurposing activities such as feasibility assessment, acquisition, exposition, and transformation.
Data quality for repurposing cannot be judged by fitness-for-use alone; accessibility, transparency, and elasticity must also be evaluated.
Design choices made during original data management, such as granularity and normalization, directly determine how repurposable the data will be later.
The framework provides a common vocabulary that unifies task-driven and data-driven repurposing, as demonstrated in the citizen-science and healthcare cases.
Repurposed data becomes a new data asset that can itself be repurposed, creating unpredictable indirect outcomes and potential competitive advantages for data-rich organizations.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A testable extension is to operationalize elasticity as the minimum number of schema additions needed to serve a new task, which would let datasets be scored for repurposability before any particular task is chosen.
The framework implies that data governance should record design decisions and collection context at the time data are created, since later exposition depends on information that may otherwise be unrecoverable.
A further testable consequence is that fine-grained, loosely categorized datasets will require fewer schema additions across a sample of new tasks; this could be checked empirically using the citizen-science evidence the paper cites.
For AI training, the definition suggests that feeding a model unaltered text is data reuse rather than repurposing, which could refocus debates about data provenance on cases where training data must be re-schematized or augmented.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Two things to know. First, this is a genuine synthesis: it gives a clean vocabulary for a messy area (reuse vs repurposing, accessibility, transparency, elasticity, data-task alignment) and connects prior work on FAIR, schema integration, and data quality into one frame. Second, its central definition has a real hole, and the hole is visible in the paper's own examples.

The new part is the definition of repurposing as schema augmentation, plus the process model (acquisition, exposition, transformation) and the use-agnostic properties. The authors are honest that this is a commentary, not an empirical test. The two cases (citizen science, AHCA) are well chosen and show how the framework can organize analysis. The research agenda in Section 5 is detailed and mostly unobjectionable.

The soft spot is load-bearing. Section 3.1 defines repurposing as use after one or more additions to the schema. But footnote 3 treats removing PII from server logs as part of repurposing — that is a deletion, not an addition. Section 3.2 lists filtering, debiasing, pivoting, and granularity alignment as transformations in repurposing; those do not necessarily add schema elements. And the AHCA case is called repurposing largely on the basis of asking new questions over existing fields, which the paper's own rule classifies as reuse. When the schema is implicit (which the paper allows), the boundary becomes unfalsifiable: the same query can be framed as reuse or repurposing depending on how you reconstruct the implicit schema. This matters because the reuse/repurposing distinction is the paper's object of study, not a side detail.

The fix is not hard: either broaden the definition to any meaningful schema change (additions, deletions, restructuring, granularity shifts) or recast the boundary as a continuum with reuse on one end and repurposing when transformation effort crosses a threshold. As written, it is a proposal that will generate useful debate, but the demarcation is not yet operational.

Also worth noting: Section 4.3 says the framework 'clearly' brings key aspects into focus, but the evidence is two illustrative cases. That is overclaiming and should be toned down.

Who this is for: anyone working on data reuse, data quality, data governance, or conceptual modeling would get value from the framework and the agenda, even if they end up modifying the definition. It deserves a serious referee — the issues are about precision, not about the paper being unserious.

Referee Report

3 major / 2 minor

Summary. The paper argues that data repurposing is a distinct and under-theorized practice in data management, defined as data use performed after making one or more additions to the schema of the original data, and contrasts it with original data use and data reuse. It develops a Data Repurposing Framework consisting of use-agnostic data properties (accessibility, transparency, elasticity), repurposing activities (feasibility assessment, task reconceptualization, data-task alignment, acquisition, exposition, transformation), actors, and environmental enablers and constraints. The framework is illustrated through two case studies: a task-driven citizen-science example involving bird survival and light/noise pollution, and a data-driven healthcare example using the Florida AHCA hospital discharge dataset. The paper concludes with a research agenda organized around the framework's components.

Significance. If the core demarcation between reuse and repurposing is stable, this is a valuable conceptual contribution for the data management community. The paper synthesizes scattered literature across scientific data reuse, data integration, and data quality, and it proposes concrete, usable constructs (especially elasticity and data-task alignment) that could organize future empirical work. The two case studies are rich and credible illustrations, particularly the citizen-science example, which clearly involves schema augmentation through the addition of light and noise attributes. The paper does not provide machine-checked proofs or quantitative validation, but for a commentary-style conceptual framework this is not disqualifying; the main significance hinges on whether the definition of repurposing is consistently applicable to the paper's own examples.

major comments (3)

[Section 3.1, footnote 3] The definition of repurposing as data use 'after making one or more additions' to the schema is contradicted by the paper's own examples. Footnote 3 describes trace data being repurposed for consumer behavior analysis after 'removing potentially identifying information,' which is a schema deletion, not an addition. Similarly, Section 3.2 lists 'filtering, debiasing, pivoting and granularity alignment' as transformations in repurposing; these operations can change the data without adding any new attribute or entity type. Under the stated rule, such uses would be classified as data reuse, yet the paper presents them as repurposing. This internal inconsistency undermines the central claim that a schema addition is the defining feature of repurposing.
[Section 3.1, paragraph on 'explicit or implicit' schema] The reuse/repurposing boundary is not operationally decidable when the schema is implicit. The paper defines a schema as an 'explicit or implicit conceptual representation,' but when the schema is implicit, whether a new task requires an addition to that schema cannot be determined from the data alone; the same task can be framed as either reuse or repurposing depending on how one chooses to make the implicit schema explicit. This matters because the paper claims that schema overlap 'makes it possible to establish the limits of repurposing.' Without a method for fixing the content of an implicit schema, the central demarcation is under-specified.
[Section 4.2 (AHCA case)] The AHCA case does not identify a specific schema addition for the repurposing examples it cites. The text argues that the data have been 'extensively repurposed' for academic research, but the uses described—querying existing fields such as patient sex, race, or physician identifiers, and filtering the data—are operations that conform to the existing schema. Under the paper's own Section 3.1 definition, these would be data reuse rather than repurposing unless the narrative specifies which external data sources (e.g., Hospital Compare) were added as new attributes or entity types. As written, the case does not demonstrate repurposing as defined, which weakens the framework's claimed applicability to a major illustrative example.

minor comments (2)

[Acknowledgements] The acknowledgements contain a typo: 'The authors wish to than' should read 'The authors wish to thank'.
[References (Norman, 2002)] In the reference list, 'Basic Books' is misspelled as 'Bsic Books'.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity; the paper is a conceptual proposal whose definition is stipulated rather than derived, and its self-citations are not load-bearing.

full rationale

The paper's central claim is an explicit definition: data repurposing is defined as data use after making one or more additions to the schema describing the original data (Section 3.1), with data reuse defined as use that the existing schema can accommodate. This is a conceptual stipulation, not an empirical derivation, so it is not circular in the sense of fitting a parameter and renaming it a prediction. The framework's activities and properties are proposed conceptually and then illustrated with two case studies; the cases are interpretations under the framework rather than tests that the framework is supposed to predict. Some self-citations appear, notably Parsons and Wand (2000) for conceptual data independence and Lukyanenko et al. (2019) for anchoring effects, but these support specific constructs and do not carry the central demarcation between reuse and repurposing. The reuse/repurposing boundary may be under-specified, especially when examples involve schema deletions or implicit schemas, but that is a conceptual ambiguity rather than circular reasoning. No equations, fitted parameters, or externally derived predictions are present, so the score is low.

Assumptions & free parameters 0 free parameters · 4 assumptions · 3 invented entities

The framework rests on stipulative definitions and posited constructs. No free parameters are fitted to data. The invented constructs (elasticity, data-task alignment, conceptual data independence) organize the domain but lack independent empirical handles, so they carry a theory-building burden rather than a falsifiable prediction.

assumptions (4)

domain assumption Data are symbols that always have an original purpose; the purpose is encapsulated in a schema that data conform to.
Section 3.1 introduces the symbol-schema view (citing Morris 1946, Alaimo and Kallinikos 2022) as the foundation for defining repurposing. This premise is not tested.
ad hoc to paper Repurposing is defined by schema augmentation: any new use that can be served by the existing schema is data reuse, not repurposing.
Section 3.1 stipulates the boundary between reuse and repurposing. This categorical definition is the paper's central postulate and is not empirically derived.
ad hoc to paper Accessibility, transparency, and elasticity are the use-agnostic properties that enable or inhibit repurposing.
Section 3.2 posits these three properties as the key dimensions for repurposability. They are introduced by the authors without a derivation or empirical justification.
domain assumption The antecedents of elasticity (data presentation, granularity, storage, governance) determine repurposability.
Section 3.2 lists these antecedents based on illustrative examples and prior work; they are plausible but not established.

invented entities (3)

Elasticity
purpose: To capture the extent to which data can be aligned with new schemas and tasks after transformation.
Introduced in Section 3.2 as a new use-agnostic property. No falsifiable measure or external benchmark is provided.
Data-task alignment
purpose: To describe the degree of match between available data and the ideal schema for a new task, guiding acquisition and transformation.
Defined in Section 3.2 as a central concept. It is a conceptual construct used to frame activities, without a quantitative operationalization.
Conceptual data independence (extended to repurposing)
purpose: To formalize the idea that data can be decoupled from their original schema, extending Codd's data independence.
Section 3.2 extends Parsons and Wand (2000) to repurposing. It is a theoretical extension, not a new measurable entity.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Understanding and Improving Data Repurposing." pith.science (2026). https://pith.science/paper/5EBTRA4O

@misc{pith2026250609073,
  author       = {Pith},
  title        = {Pith review of: Understanding and Improving Data Repurposing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5EBTRA4O}},
  note         = {Machine review of arXiv:2506.09073}
}

read the original abstract

We live in an age of unprecedented opportunities to use existing data for tasks not anticipated when those data were collected, resulting in widespread data repurposing. This commentary defines and maps the scope of data repurposing to highlight its importance for organizations and society and the need to study data repurposing as a frontier of data management. We explain how repurposing differs from original data use and data reuse and then develop a framework for data repurposing consisting of concepts and activities for adapting existing data to new tasks. The framework and its implications are illustrated using two examples of repurposing, one in healthcare and one in citizen science. We conclude by suggesting opportunities for research to better understand data repurposing and enable more effective data repurposing practices.

Figures

Figures reproduced from arXiv: 2506.09073 by the authors.

**Figure 1.** Data Repurposing vs Original Data Use The extent of data-task alignment depends on the similarity of the ideal schema to the original one, making it possible to establish the limits of repurposing. Repurposing is possible as long as there is an overlap in attributes between the original schema (Schema 1 in [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

19 extracted references · 18 canonical work pages

[1]

As a result, some knowledge is driven by shared understanding of factors such as why data is needed and who created it

icitly documenting every aspect of data. As a result, some knowledge is driven by shared understanding of factors such as why data is needed and who created it. This is epitomized by agile development (Johnson, 2020), which is characterized by continuous and close interaction between users and software developers such that systems, including data structur...

work page 2020
[3]

Hence, repurposing is affected by policies, guidelines and ideas that constantly evolve (Krämer & Shekhar, 2025)

Repurposing enablers and constraints Enablers & Constraints Task Original Data Alignment Outcomes People Skills and knowledge Awareness of sources of data for a task Ability to understand existing data Ability to acquire, expose, transform data Ability to realize analysis into change Motivation Interest in identifying questions to be addressed Interest in...

work page 2025
[6]

(Re-)Conceptualization

is a valuable aspect of data management for repurposing. (Re-)Conceptualization. There is a large body of work on conceptualizing data for specific tasks (for reviews, see Chua et al., 2022; Recker et al., 2021; Storey et al., 2023; Wand & Weber, 2002). Traditional approaches to conceptualization assume the resulting data will align well with the requirem...

work page 2021
[8]

data swamp

or automated machine learning (Larsen & Becker, 2020). Danger exists when transformations are opaque and it is unclear what decisions were taken when data were transformed or what information was lost in the process. Repurposing data without understanding how it has been transformed can lead to inappropriate conclusions and actions. Despite the rise in th...

work page 2020
[9]

P., & Larson, J

Sheth, A. P., & Larson, J. A. (1990). Federated database systems for managing distributed, heterogeneous, and autonomous databases. ACM Computing Surveys (CSUR), 22(3), 183–

work page 1990
[10]

– can both enable and constrain repurposing and its impacts need to be better understood. To answer certain questions in a truly innovative way, one must think outside the box and 9 https://www.nytimes.com/2023/12/27/business/media/new-york-times-open-ai-microsoft-lawsuit.html MIS Quarterly Data Repurposing 26 find sources that on the surface have nothing...

work page 2025
[17]

MIS Quarterly Data Repurposing 29 Abraham, R., Schneider, J., & Vom Brocke, J. (2019). Data governance: A conceptual framework, structured review, and research agenda. International Journal of Information Management, 49, 424–438. Alaimo, C., & Kallinikos, J. (2022). Organizations decentered: Data objects, technology and knowledge. Organization Science, 33...

work page 2019
[19]

Woodall, P. (2017). The data repurposing challenge: New pressures from data analytics. Journal of Data and Information Quality (JDIQ), 8(3–4), 1–4. Wu, C.-J., Raghavendra, R., Gupta, U., Acun, B., Ardalani, N., Maeng, K., Chang, G., Aga, F., Huang, J., & Bai, C. (2022). Sustainable AI: Environmental implications, challenges and opportunities. Proceedings ...

work page 2017

Show all 19 references

[30]

Vermicelli, S., Cricelli, L., & Grimaldi, M. (2021). How can crowdsourcing help tackle the COVID‐19 pandemic? An explorative overview of innovative collaborative practices. R&D Management, 51(2), 183–194. Wand, Y., & Weber, R. (2002). Research commentary: Information systems a...

2021
[66]

The academic community can do more to understand such issues and improve existing practices

suggest: “[r]eusing and repurposed data requires reimagining their roles in building their organizational capabilities to fuse such operations with established organizational and knowledge functions and practices.” Reflecting on issues related to getting most out data in the c...

2024
[236]

Sleeper, R. (2020). Innovative Tableau: 100 More Tips, Tutorials, and Strategies. O’Reilly Media. Spaccapietra, S., & Parent, C. (1994). View integration: A step forward in solving structural conflicts. IEEE Transactions on Knowledge and Data Engineering, 6(2), 258–274. Storey...

2020
[429]

L., Kallinikos, J., & Monteiro, E

Aaltonen, A., Alaimo, C., Parmiggiani, E., Stelmaszak, M., Jarvenpaa, S. L., Kallinikos, J., & Monteiro, E. (2023). What is Missing from Research on Data in Information Systems? Insights from the Inaugural Workshop on Data Research. Communications of the Association for Inform...

2023
[1988]

chief health policy and planning entity

These data have been extensively repurposed for academic research (e.g., Burke et al., 2007; Greenwood & Agarwal, 2015), as well as for policy purposes. 7 Even so, we are unaware of a systematic evaluation of the suitability of the AHCA data for repurposing. By applying the Da...

2007
[2005]

As there has been a proliferation of general ontologies, systematic investigation of their potential to support data exposition for repurposing is needed

as a basis for developing and validating domain specific ontologies to increase transparency. As there has been a proliferation of general ontologies, systematic investigation of their potential to support data exposition for repurposing is needed. Often, little is known about...

2021
[2014]

Furthermore, many practices are new, such as the generation of synthetic data by AI for repurposing as training data for new tasks

but not how data was transformed. Furthermore, many practices are new, such as the generation of synthetic data by AI for repurposing as training data for new tasks. Given the impact of transformation on repurposed data, an important opportunity is developing approaches for do...

2021
[2021]

economic moat

that can be bought or rented and further repurposed by combining with other data to draw additional insights. MIS Quarterly Data Repurposing 14 Repurposing per se concludes with the completion, abandonment, or redefinition of the task following the analysis of the data. We con...

2002
[2024]

The problem is not trivial, as the value of repurposing may only be evident after costly data acquisition and transformation (Grande et al., 2020; Parmiggiani et al., 2022)

lack a feasibility analysis stage for data repurposing. The problem is not trivial, as the value of repurposing may only be evident after costly data acquisition and transformation (Grande et al., 2020; Parmiggiani et al., 2022). Hence, an important research challenge is estim...

2020
[2025]

https://www.gartner.com/en/newsroom/press- releases/2021-05-19-gartner-says-70-percent-of-organizations-will-shift-their-focus-from- big-to-small-and-wide-data-by-2025 Gorelik, A

Gartner. https://www.gartner.com/en/newsroom/press- releases/2021-05-19-gartner-says-70-percent-of-organizations-will-shift-their-focus-from- big-to-small-and-wide-data-by-2025 Gorelik, A. (2019). The enterprise big data lake: Delivering the promise of big data and data scienc...

2019
[2303]

Digital First

MIS Quarterly Data Repurposing 31 Greenwood, B. N., Agarwal, R., Agarwal, R., & Gopal, A. (2017). The when and why of abandonment: The role of organizational differences in medical technology life cycles. Management Science, 63(9), 2948–2966. Guizzardi, G. (2005). Ontological ...

2017 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

As a result, some knowledge is driven by shared understanding of factors such as why data is needed and who created it

icitly documenting every aspect of data. As a result, some knowledge is driven by shared understanding of factors such as why data is needed and who created it. This is epitomized by agile development (Johnson, 2020), which is characterized by continuous and close interaction between users and software developers such that systems, including data structur...

work page 2020

[2] [3]

Hence, repurposing is affected by policies, guidelines and ideas that constantly evolve (Krämer & Shekhar, 2025)

Repurposing enablers and constraints Enablers & Constraints Task Original Data Alignment Outcomes People Skills and knowledge Awareness of sources of data for a task Ability to understand existing data Ability to acquire, expose, transform data Ability to realize analysis into change Motivation Interest in identifying questions to be addressed Interest in...

work page 2025

[3] [6]

(Re-)Conceptualization

is a valuable aspect of data management for repurposing. (Re-)Conceptualization. There is a large body of work on conceptualizing data for specific tasks (for reviews, see Chua et al., 2022; Recker et al., 2021; Storey et al., 2023; Wand & Weber, 2002). Traditional approaches to conceptualization assume the resulting data will align well with the requirem...

work page 2021

[4] [8]

data swamp

or automated machine learning (Larsen & Becker, 2020). Danger exists when transformations are opaque and it is unclear what decisions were taken when data were transformed or what information was lost in the process. Repurposing data without understanding how it has been transformed can lead to inappropriate conclusions and actions. Despite the rise in th...

work page 2020

[5] [9]

P., & Larson, J

Sheth, A. P., & Larson, J. A. (1990). Federated database systems for managing distributed, heterogeneous, and autonomous databases. ACM Computing Surveys (CSUR), 22(3), 183–

work page 1990

[6] [10]

– can both enable and constrain repurposing and its impacts need to be better understood. To answer certain questions in a truly innovative way, one must think outside the box and 9 https://www.nytimes.com/2023/12/27/business/media/new-york-times-open-ai-microsoft-lawsuit.html MIS Quarterly Data Repurposing 26 find sources that on the surface have nothing...

work page 2025

[7] [17]

MIS Quarterly Data Repurposing 29 Abraham, R., Schneider, J., & Vom Brocke, J. (2019). Data governance: A conceptual framework, structured review, and research agenda. International Journal of Information Management, 49, 424–438. Alaimo, C., & Kallinikos, J. (2022). Organizations decentered: Data objects, technology and knowledge. Organization Science, 33...

work page 2019

[8] [19]

Woodall, P. (2017). The data repurposing challenge: New pressures from data analytics. Journal of Data and Information Quality (JDIQ), 8(3–4), 1–4. Wu, C.-J., Raghavendra, R., Gupta, U., Acun, B., Ardalani, N., Maeng, K., Chang, G., Aga, F., Huang, J., & Bai, C. (2022). Sustainable AI: Environmental implications, challenges and opportunities. Proceedings ...

work page 2017