In order for your dataset to be more easily accessible to users, it often makes sense to add in additional data from other sources. For example, in a dataset of European tender award notices the CPV (common procurement vocabulary) code for each contract may be given. While the code itself is not very helpful, adding in the EU's descriptions of each code will give visitors of your dataset a clearer idea of what the money is being spent on.
Here are some ideas for extending your data:
- Classification codes (such as the CPV mentioned above, COFOG, the classification of functions of government, or DAC codes in development cooperation) often occur without much explanation in the source data. Try to find a code sheet that assigns names and hierarchies to these codes, so that users can understand what they mean and how they relate to each other. Sometimes classification schemes are maintained only within an organization, so it may be worthwhile sending a freedom of information request to get access to code sheets.
- Companies and other organizations are often identified only by their name and possibly their VAT code. Companies databases such as OpenCorporates.com offer more information on each company that you can merge into your dataset. Such information may include its geographic location, industry classifiers or board members.
- Sometimes, separate databases are kept on project finance and results reporting. You might be looking at a dataset that only contains grant amounts and general information, when data on disbursement, progress or more descriptive texts are available elsewhere. Try to find a linking element between the two databases (e.g. a project ID) and to merge the two databases.
- Geographic information can be extended using various web services, such as OpenStreetMap. Using such services you may be able to infer state and country names from a given city label or even to include geo-coordinates for projects.
Some of these proposals require you to merge two or more spreadsheets. While this can be done in newer versions of Microsoft Excel and LibreOffice, data wrangling tools such as Google Refine (and its sadly underdocumented cell.cross() function) and Google Fusion Tables can be very useful in this regard. Depending on the amount of processing that you need to do, it may also make sense to import your data into a relational database such as SQLite. While there are visual frontends for these databases, using them to merge datasets will ususally require you to be familiar with SQL.