Status of quality evaluation

[Mika Saarenmaki]

We have not been able to finish the evaluation of data quality for our database and would like to know about the situation in other countries. We have prioritized evaluation of recent analytical data and use the EuroFIR procedure (1). Excel spread sheets are still used for the evaluation. We sometimes get quite high score for old data when we have all the background information. This has been criticized, especially when we have few samples. We have not started work on complimentary data (2). Are there any problems with the quality assessment in other countries  and how much have you implemented from the EuroFIR Guidelines? Ref: (1) Guidelines for quality index attribution to original data from scientific literature or reports for EuroFIR data interchange. EuroFIR workpackage 1.3, task group 4. (2) Guidelines for quality index attribution to complimentary data for EuroFIR data interchange. D1.3.26. 

[[email protected]]

 In September 2012, we finished collecting Slovenian food composition data on plants. Now, we have started the next phase, evaluating the data quality. As there are no Slovenian recommendations for quality assessment, we stick to the EuroFIR guidelines.

[Anna Giertlova]

We have evaluated some new scientific data by using EuroFIR guidelines for attribution QI to original data from scientific literature. However, we have opposite experience. The final score or QI of evaluated scientific data was not the highest, as we expected, due to the lack of information (e.g. we have had problem to identify food or better description of analysed food was missing).

[susanne westenbrink]

In NL we evaluate the quality of data before taking the decision to use it in our database or not. These procedures are integrated in our work flow and match with the way our DBMS system works. However we do not give a score to individual values or data sources.

We did not yet start to apply the EuroFIR QI attribution. Since we need to maintain this outside our DBMS this is not very practical. As soon as we have a way to integrate QI assessment within our DBMS I hope we can make a start to QI all new analytical values.

EurFIR NEXUS compilers will soon be asked to cooperate in one of the NEXUS workpackages to work on quality indexing of some selected values.

[Marine Oseredczuk]

All our data are currently assessed with a specific data quality assessment system that was internally developped at Anses many years ago. There are indeed 4 different sets of criteria for assessment, with 4 corresponding rating methods. One set of criteria and associated rating method is for data from scientif litterature or reports, the others are for calculation, for industry data and for data from other tables or databases. Most of the principles used in our specific quality assessment system are close to those used in the EuroFIR proposal, but the EuroFIR proposal is more comprehensive.

Switching from our current system to the EuroFIR one will be challenging, but it must be possible. We could try that after the EuroFIR data quality system is finalized.

Regarding feedbacks we have had on the quality scores we assign, they were generally positive. When collecting data directly from data producers, we try to present the data quality assessment system we use to data producers. This presentation include description of our needs as food data compilers and explainations on how the compiled data we will produce from the collected data will be used. It is the opportunity to have interesting exchanges with data producers and to share experiences and knowledge from both sides. Then, the result of the data quality assessment we made is well understood and rarely criticized from data producers we have met. Therefore, although data quality assessment is a long task and requires communication with the various groups involved (data producers, data users), I think that what we have gained a lot from it.

Greetings to all compilers!

Marine

[Author]

Posts