How to use table Linkbase information to facilitate XBRL report analysis Jean-Paul Daisomont 1 and Denis Robinet 2 1 acsone, Brussels, Belgium jean-paul.daisomont@acsone.eu 2 Haute École Robert Schuman, Wallonia, Belgium Abstract. This paper explores the technical challenges and difficulties to cover table Linkbase and dimensions management in case of open axes. These charac- teristics are mainly present in Solvency II taxonomy, and, to a lesser extent, in CRD IV taxonomies. This exploration started last year when we discovered fact description given by OIM specification. Of course, OIM target is to simplify XBRL reports formatting and exchange. Fact description in OIM highlights clearly the relationship between facts and dimensional information. From our side, this triggers a reflection over the exploitation of dimensional descriptions in the context of data dissemination. In other words, it concerns the way to en- rich fact description with taxonomy information to help data analysis. One of the main expected qualities is to avoid specific development related to the specificity of the XBRL reports. This requires working on the abstract defi- nition of XBRL taxonomies. In this paper, we will first describe which source of information will be han- dled, how it will be processed and designed into a database schema. This study is currently relevant for European taxonomies, using table Linkbase specifica- tions, as specified by EIOPA and EBA. . 1 What sources of information can be exploited? Three types of information seem useful for data analysis: (1) aspects of the facts, (2) table characteristics expressed in X-Y-Z axes and in table structure information, and (3) taxonomy specification. 1.1 Aspects of the facts Aspects are immediately available within XBRL reports; simple parsing of XBRL report is enough to retrieve this type of information. Labels can be retrieved from taxonomy specifications. Here after, we illustrate for a fact from Solvency II report in OIM CSV format: 2 Fig. 1. Extract from report in CSV OIM format Here after, the same fact with its XBRL properties (on the left side) and its table presentation (on the right side). Fig. 2. XBRL properties of a fact from Solvency II report in the case of not open table. 1.2 Table characteristics of the facts Facts in reports are only defined by its aspects. For analysis, information over table structure is also relevant. Three types of information over table structure are relevant: • Ventilation of the dimensions over axes; • Description of table structure in terms of axes, breakdowns and structural nodes; • Labels related to concepts, dimensions, members and table structure Ventilation of dimensions over axes. Here is the ventilation of the fact dimen- sions over the axis of the fact from table S.05.02.01.01; members of the dimensions are defined by the following table Linkbase. 3 Dimensions Members X-axis Y-axis Z-axis S2c_dim:BL S2c_LB:x79 S2c_dim:DI S2c_DI:x5 S2c_dim:EE S2c_GA:x24 S2c_dim:IZ S2c_RT:x1 S2c_dim:TB S2c_LB:x107 S2c_dim:VG S2c_AM:x84 This is a simple case because axes of table S.05.02.01.01 are closed. Solvency II taxonomy use quite frequently open axes; the most sophisticated situation is present in table S.22.06.01.04 where all the axes X, Y and Z are open. Here follows an illustration of this case. Fig. 3. XBRL properties of a fact from Solvency II report in the case of open table. The ventilation of the fact dimensions over the axis is the following: Dimensions Members X-axis Y-axis Z-axis S2c_dim:BL S2c_LB:x65 4 S2c_dim:EE S2c_GA:x74 S2c_dim:LG S2c_GA:DK S2c_dim:OC S2c_CU:EUR S2c_dim:RC S2c_CU:x4 S2c_dim:VG S2c_AM:x80 S2c_dim:VL S2c_VM:x5 In this case, member definitions are coming from the table Linkbase and also from the report fact. In that case, the table Linkbase specifies the allowed set of values. The third interesting case occurs when the member is defined by the entity in the report. Table S.09.01.01.01 from Solvency II illustrates it; one dimension, XD, is attached to the fact and to Y-axis. Dimensions Members X-axis Y-axis S2c_dim:XD 1 Fig. 4. : XBRL properties of a fact in the case of Y-axis open table with dimension member defined by the entity. Table structure in terms of axes, breakdowns and structural nodes. In addition to the axes description, information over breakdowns and structural nodes are also useful to use the table structure. Breakdowns and structural nodes are defined in table Linkbase specifications. 5 Labels. Data without labels is not useful; labels over table structure can be re- trieved within table Linkbase specifications. 2 How can it be processed? Previous chapter draws the list of wishable information over table structure that can be useful for data analysis. This chapter will describe the organisation of the data (database schema) that we suggest storing reports. Two sets of data/processing have to be distinct: • Data/processing related to the taxonomy • Data/processing related to the report The processing that we have implemented is based on the technical environment of Arelle project; it means that its processing works on taxonomy specifications only and does not include taxonomy dependent processing. 2.1 Data model for the taxonomy The information that we retrieve from taxonomy specifications are: taxonomy ver- sion, table, breakdown (orientation means axis), structural node, member, domain, dimension and relationship between structural node and member. 6 Fig. 5. Data model for the taxonomy. 2.2 Data model of the report Information specific to the report are: report, fact, context, user dimension and rela- tionship between user dimension and context. A lot of links exist between report and taxonomy data; the main are the link between: • fact and structural node. • user dimension and taxonomy dimension. • context and taxonomy dimension and taxonomy member. • fact and taxonomy member. 7 Here after the data model combining report and taxonomy tables. We use colours to distinct tables related to taxonomy from tables linked to reports. Fig. 6. Data model for taxonomy and report. 3 Conclusion In Europe, the learning processes over new regulations are based on table description of reports such as QRT. Unfortunately, the excellent OIM specification does include not information over that point of view. Our approach is to enrich the OIM with X-axis, Y-axis and Z-axis information; that table presentation information will mainly be useful to load XBRL reports into data- base and data warehouse. This paper tries to show which table information can be extracted from taxonomies and exploited for data analysis. It assumes that table Linkbase specifications are available. The approach presents some limitations: • Multilingual information is not yet managed; it can be useful for the French RAN taxonomy, the Dutch Specific templates (Verzekeraars Nationale StatenTaxon- omy), the Single Resolution Board. • It does not yet support presentation Linkbase taxonomies. • Member hierarchies are not explored.