Improve support for numeric datatypes

[jaw111]

Community Note

• Please use a 👍 reaction to provide a +1/vote. This helps the community and maintainers prioritize this request.
• If you are interested in working on this issue or have submitted a pull request, please leave a comment.

Graph Notebook Version (and Graph Database and Version used if applicable)
4.6.0 (Amazon Neptune 1.2.1.0)

Is your feature request related to a problem? Please describe.
When I use the %%sparql magic with --store-to and --store-format parameters to store the results of a SPARQL SELECT query to a dataframe, all the values in the dataframe are handled as strings, whilst the RDF literals have standard XSD datatypes like xsd:integer or xsd:decimal.

Describe the solution you'd like
When processing the results of a SELECT query, add handling for standard XSD datatypes that can be handled as numerics in a dataframe.

Also add error handling if the lexical value cannot be interpreted according to the datatype. For example a typed literal like "example"^^<http://www.w3.org/2001/XMLSchema#integer> may occur in RDF data, but the value cannot be interpreted as an integer. Could fallback to string in such cases.

Additional context
Workaround is to add a post-processing step to convert values from string to numeric for each variable/field where we expect numeric values:

df.x = pd.to_numeric(df.x, errors='coerce')