--- title: "New Requirements" output: rmarkdown::html_vignette vignette: > %\VignetteIndexEntry{New Requirements} %\VignetteEngine{knitr::rmarkdown} %\VignetteEncoding{UTF-8} --- ```{r vignettesetup, include = FALSE} knitr::opts_chunk$set( collapse = TRUE, comment = "#>" ) ``` ```{r setup} library(dataset) ``` The first CRAN release and rOpenSci review brought very useful experience and feedback. The dataset package had been defined with a very broad requirement. While the very general requirement setting has advantages, a clear disadvantage is that without a specific use case, it is difficult to raise enough user and co-developer interest. In line with some of the legitimate criticism of version 0.1.0—0.3.1, I envision a dataset package that has more specific inheritance packages that work with datasets in a more specific use case. It is probably too wide of a claim to create a package that will bring the base R data.frame object in line with any disciplinary requirements of datasets. The original concept closely followed the SDMX (statistical) dataset definition to the extent that one reviewer recommended the `datacube` name for the package. In light of some further use experiences, this is a valid criticism because datasets used in digital humanities, for example, have slightly different specification needs. R is primarily a statistical environment and language; therefore, broad conformity with the SDMX statistical metadata standards is desirable. However, the dataset package should remain generic enough to support non-statistical datasets. Along these lines, the current aim is to triangulate three packages: - A `datacube` package, which follows more closely the SDMX definition of a dataset and the more general, multi-dimensional datacube definition - A `wbdataset` package, which follows more closely the Wikibase Data Model that is increasingly used for digital collections management and other scenarios for statistically not aggregated datasets. - A `dataset` package that is sufficiently generic that both the `datacube` and the `wbdataset` package rely on it as a joint dependency. Therefore, the plan is to relax some SMDX definitions of datasets that are not very useful in non-statistical applications. Such functionality can be removed from a later-developed `datacube` package. At the same time, I would like to co-develop the dataset package with the `wbdataset` package because the Wikibase Data Model is a very well-defined semantic data model that could potentially create a large enough user base and use case for the entire project. Another important lesson was that the first version of the dataset package wanted to be so generally usable that it aimed for compatibility for base R data.frames, the tidyverse tibble modernisations of such data frames, and the data.table objects, which have their own user base and dependencies in many statistical applications. While such broad appeal and ambition should not excluded for the future, it would be a too significant undertaking to ensure that all functionality works with data.frames, tibble and data.tables. Whenever this is possible, this should remain so, but new developments should only follow the modern tidyverse tibbles. ## Tidy can be tidier Most R-user data scientists are familiar with the term tidy data. It is data ready for tabular analysis and easy to understand. It places measurements (variables) in columns, dividing each observation into a row. Column names convey the meaning of the variable. You should wrangle your data in a tidy format because that is the prerequisite of most statistical analysis or visualisation with various algorithms, and it gives a logical place to each data point. Making your data tidy is like hoovering your room. It will make your environment neat, but if you leave for months, it will be full of dustballs, insects and spiders, and you will not remember where you left your slippers. To allow us to return to work after months or years or to pass on our work to others, we need to provide more meaning (semantics) about a tidy dataset. Consider the following simple data.frame: ```{r example-ambiguity} data.frame( geo = c("LI", "SM"), CPI = c("0.8", "0.9"), GNI = c("8976", "9672") ) ``` This dataset is tidy. But it certainly could be improved! - `geo`: you may figure out that geo means something to do with geography (maybe countries, "LI" standing for Lichtenstein and "SM" for San Marino), but even that, who knows? If you add Greece, should use use "EL" like Eurostat or "GR" like the World Bank? - `CPI` can stand for "consumer price index" or "corruption perceptions index". Or anything else! - `GNI` can stand for "Gross National Income" and "Global Nutrition Index", often mentioned in the same context. If there are numbers, do they mean a physical quantity or a currency? Is the currency dollars or euros? Adding such definitions to the dataset makes them far more accessible, reusable, and even findable. Many people may be looking for GNI in US dollars; using a currency cue for searching may bring in a new user. Most researchers are familiar with the semantic needs to make a dataset findable; they may know how libraries or scientific repositories store authorship data or titles. Our package handles these types of metadata, too, but goes into the semantics of what's inside the dataset. What does each column mean? What is the identity of the observations in the rows (if they are not anonymous)? Adding such information in a machine-readable way allows far greater findability because people can search for data expressed in euros or dollars only or any datasets that use CPI in the sense of the Corruption Perception Index instead of finding thousands of search hits for CPI as consumer price inflation. Further enriching the semantics of an already tidy dataset can increase findability and reusability. Even the same user may struggle months or years later with a saved tidy dataset that is missing the definitions or variables of observations, or the unit of measure. ## Exchange and extend Our dataset package focuses on an R user who would like to work on maintaining, extending or improving the same dataset for many months or years, potentially with several data managers and data sources under joint stewardship. - Extending the rows means adding new observations which follow clear semantics on observation identification. In the previous example, if the observation units are countries, we need to make it clear how a user can add another country in a new row; and what happens if the same country is added with different measurements in different points of time. If you use two-letter country codes like "SM" and "LI", then adding Brazil should use "BR" and not "BRA". Eurostat abbreviates Greece as "EL", but IMF uses the ISO-standard "GR", so the taxonomy of abbreviations should be available. - Extending the columns means adding new attributes or measured values for each observational unit in a way that they increase the usability of the dataset (and do not confuse it.) If you have a GDP and a population column, you can calculate a GDP/Capita, but only if you divide the GDP and the population from the same time period, and GDP measured in the same currency unit. Often it is important to know the exact formula of a new, mutated variable; for example, "GDP_mean" can mean a geometric or arithmetic or harmonic mean, too! The magic abbreviation FAIR data stands for findable, accessible, interoperable and reusable; F, A, and R mainly boil down to adding the correct metadata to the data. The European Interoperability Framework as a standard describes four layers of interoperability, so it is a far more elusive concept. In this standard, interoperability has a legal, organisational, semantic, and technical layers. Technical interoperability can be increased by releasing the dataset in a system-independent format such as JSON. Legal interoperability can be increased by adding a standard re-use rights statement and authorship information about the dataset itself to the release. But considering the R user's workflows (organisational aspect) and communicating it in a meaningful way (semantic aspect) ensures that a new user can effectively improve the dataset (similarly to adding to the code base of an R package), or use the dataset in an entirely different job. Making sure that the "GNI" is about nutrition and not national income, the "CPI" is about inflation and not corruption is essential to place your dataset into a public health or a macroeconomic analysis workflow. ## Placing the `dataset` package into the R ecosystem Many R packages try to help FAIR qualities with different mindsets and tools. The dataset package differs from them and can work with many of them synergistically, magnifying the value added by other metadata packages. - [x] The `frictionless` R package family (), the rOpenSci `dataspice` and the rOpenSci ` rdflib allow you to serialise your data into a semantically rich format for different purposes. - [x] The `dataspice` creates HTML files that are easy to find on the internet with search engines; it relies heavily on the lightweight semantic ontology of Schema.org, which ensures that websites understand certain metadata the same way. - [x] The `rdlib` package allows annotating with the World Wide Web RDF standard markup any three-column tidy dataset and saving it to every file format the W3C consortium defined for releasing interoperable data. It provides the widest interoperability in terms of importing and exporting data, but you can make the most out of it if you are familiar with ontologies and advanced metadata modelling; it is also restrictive in terms of using long-form data. The `frictionless` family is a bit of an alternative to `dataspace` and `rdflib`: it supports one serialisation format, JSON, it heavily relies on Schema.org, and it provides interoperability within the frictionless ecosystem. The [frictionless R package](https://github.com/frictionlessdata/frictionless-r). It works well in the ecosystems of the Open Knowledge Foundations data ecosystem. These packages help release or publish data and make it more findable and reusable for new users. Our `dataset` package can be seen as boosting the performance or usability of these packages, but it has a different interoperability platform in sight, the exchange of datasets themselves either among R users or with platforms that publish datasets themselves and not text about the contents of the dataset. While the aforementioned packages mainly help with exporting datasets from R to a different format, and a different system, and perhaps outside of the statistical community in HTML, JSON, or XML. The dataset package focuses on the R user's data.frames, tibbles, potentially saved as rds or Rda files, and to collect and add as much well-designed metadata for interoperability and reuse as possible; of course, such metadata can be passed on to `dataspice`, `rdflib` or `frictionless` when leaving the R ecosytem. The `dataset` package does not want to make these packages explicit dependencies, so it offers NQuad serialisation besides CSV to export data outside of the R system; these Nquads can be translated to other standard formats with `rdflib`. We also recommend the use of `rdflib` for heavy-duty serialisation; but we offer a lightweight solution to avoid a hard dependency. The `rdflib` package is superior in exporting data from the R ecosystem to another statistical or other scientific use with the W3C standard RDF markup. For communicating among R users, it is an unnecessary translator that requires quiet advanced understanding of the RDF metadata language. It is unnecessary within the R ecosystem if we can convey the same information within an `.rds` or `.Rda` file. Similarly, if you want to communicate the contents of your dataset to non-R users or using visualisations and text, the `dataspice` package is a very good way to that. However, it is inferior for passing on an R object to another user, because it requires translation at exporting and re-translation at importing back to R; furthermore, a lot can be lost in translation as it hardcodes the Schema.org vocabulary that is not designed for statistical use, but for web publishing. ## A focus on statistical users and statistical datasets The `dataset` package focuses on the R ecosystem and R data.frames, and the R user who downloads data from different data sources like Eurostat, IMF, and World Bank and needs to join such data meaningful, add further derived own work, and save the result in .rda or .rds files. The `dataspice` and `frictionless` ecosystem focuses on data users; they are organised around Schema.org, which helps web content, such as a description of a dataset findable on a usual browser search. The `dataset` package focuses on data producers and is optimised for web queries on the web of data layer of the internet that connects databases. Therefore our default settings and the function interface focuses on the language of SDMX, the statistical data and metadata exchange standard, and DCTERMS, which is the main language of libraries and repositories. (We also support DataCite that is often preferred in European data repositories instead of the more generic DCTERMS.) These standards are not so good to find a description and a visualisation of a dataset with an web text search (Schema.org was designed for that purpose), but they are far superior for communicating the contents of a dataset to an other data producer, or somebody who may have a matching component to the contents of your dataset. Our package is not intended to those who want to effectively communicate about the interpretation of a dataset to fellow biologists or economist, but to users who want to communicate the contents of the dataset to fellow data collectors or producers. We are planning a family of dataset packages that help you with data exchanges (receiving, joining, republishing) in different data interoperability scenarios. h dataset package mainly changes the attributes system of R objects. e R users actively use the attributes; we created a reference function with sensible defaults for most statistical and open science standards. - The planned `datacube` package will extend the `dataset` package to the full SDMX Datacube specification and allow you to exchange information or publish statistical data that are fully interoperable with Eurostat or IMF datasets on statistical data repositories, for example, on the EU Open Data Portal. On such exchanges we mean that the users download both the GDP dataset and the population dataset and send back their own GDP/population calculations in a new or updated dataset. - The `wbdataset` package enables the exchange and publication of data on Wikidata or Wikibase instances, which offer the widest-used open data exchange platform for non-statistical data. It adds the main elements of the Wikibase Data Model that allow you to exchange information or publish on many open graphs (platforms that will enable connecting interoperable data sources.) A likely exchange scenario is that you download the biographical data of famous singers from a country and add back to some biographies that were not there, but you have collected them from non-interoperable sources. - A similar extension to the `frictionless` ecosystem is possible, too; should there be a user demand to improve the R-side data production workflow of data primarily intended into this ecosystem. ## New requirement settings The new dataset package would be streamlined to provide a tidier version of the [tidy data definition](https://cran.r-project.org/package=tidyr/vignettes/tidy-data.html). "Tidy datasets provide a standardised way to link the structure of a dataset (its physical layout) with its semantics (its meaning)." The aim of the dataset package is to improve the semantic infrastructure of tidy datasets beyond the current capabilities of the tidyverse packages, relaxing the exclusive use of the semantic definitions of the SDMX statistical metadata standards. The dataset package and its dependencies (or, more specifically, new extensions) will only work with tidy datasets. In the context of tidy data, "A dataset is a collection of values, usually either numbers (if quantitative) or strings (if qualitative). Values are organised in two ways. Every value belongs to a variable and an observation." Our aim is to improve the semantic description of the observations and the variables to the extent that they can be serialised to any standard format that follows the RDF W3C metadata annotation standard. Such an improvement stipulates the following requirements: - Every column must have a machine-readable name and at least one natural language label. It must contain as metadata the semantic, machine-readable definition of the variable as a URI. - Every observation must contain the machine-readable definition of the observational unit. - Whenever possible, such an observation should also be supported with a formal persistent identifier such as a URI, but this is not a strict requirement. For example, when working with this small dataset: ```{r smallcountrydataset} options(sciphen = 4) small_country_dataset <- dataset_df( country_name = defined( c("AD", "LI"), concept = "http://data.europa.eu/bna/c_6c2bb82d", namespace = "https://www.geonames.org/countries/$1/" ), gdp = defined( c(3897, 7365), label = "Gross Domestic Product", unit = "million dollars", concept = "http://data.europa.eu/83i/aa/GDP" ), population = defined( c(77543, 40015), label = "Population", concept = "http://data.europa.eu/bna/c_f2b50efd" ), dataset_bibentry = dublincore( creator = person(given = "Jane", family = "Doe"), title = "Small Country Dataset", publisher = "Reprex" ) ) ``` If you want to work with the data in the data.frame, you must know if the GDP is expressed in euros, dollars, thousands, millions or billions, in order to get to a meaningful GDP/capita figure. ```{r summaryhavenunit} summary(small_country_dataset$gdp) ``` Because the GDP is expressed in millions of dollars, we have to adjust the nominator to dollars to get the GDP per capita express in *dollars*. ```{r percapita} small_country_dataset$gdp_capita <- defined( small_country_dataset$gdp * 1e6 / small_country_dataset$population, unit = "dollar", label = "GDP Per Capita" ) ``` The `var_unit(small_country_dataset$gdp_capita)` command returns `r var_unit(small_country_dataset$gdp_capita)` and `var_label(small_country_dataset$gdp_capita)` returns `r var_label(small_country_dataset$gdp_capita)`. In this case, - we must provide a machine-readable definition of GDP, population, and GDP/capita, to improve the semantics of the columns (variables) - we must provide a machine-readable definition of the observational units of Lichtenstein and Andorra (with the optional requirement of doing so with a PID.) The metadata definitions of SDMX or open science are helpful, because they differentiate between metadata statements about the entire dataset as a structured information source, and metadata related to specific row/column intersections, where only the row, the column, and potentially the intersection cell value needs a formal definition. For example, we may want to add as metadata to the entire dataset the time frame, the usage rights, the creator. We must remain aware that in the absence of an all-encompassing, general ontology of all concepts that may go into a dataset, all datasets will be used in some kind of disciplinary context. A dataset about GDP or population will be read in a socio-economic context; a dataset containing data from sound recordings of musical works will use definitions from a musicological context. While music from Andorra may share some semantic definitions of our tiny socio-economic dataset that observes Andorra's GDP and population, it is unlikely that the observation and variable definition will lack a broad context. ```{r smallmusiciandataset1} small_country_musicians <- data.frame( qid = c("Q275912", "Q116196078"), artist_name = defined( c("Marta Roure", "wavvyboi"), concept = "https://www.wikidata.org/wiki/Property:P2093" ), location = defined( c("Andorra", "Lichtenstein"), concept = "https://www.wikidata.org/wiki/Property:P276" ), date_of_birth = defined( c(as.Date("1981-01-16"), as.Date("1998-04-28")), concept = "https://www.wikidata.org/wiki/Property:P569" ) ) small_country_musicians$age <- defined( 2024 - as.integer(substr(as.character(small_country_musicians$date_of_birth), 1, 4)), label = "Age in 2024" ) ``` Specifically, our first dataset's general semantic context is about countries as observational units and statistical variables related to them, while the second contains individual musical recordings and observed musicological values of such recordings, a general property of the dataset that can be applied to the entire dataset. Similarly, important provenance information, authorship, use rights, and so on, apply to the dataset, not to its specific cells. ```{r summarymusicians} summary(small_country_musicians) ``` The current version of dataset made efforts to record such information with applying two globally used metadata standards, the Dublin Core terms used by most librarians and perhaps the majority of scientific data repositories, and the DataCite standard preferred by some European repositories (which is more specific to data publication than generally all kinds of publication like the Dublin Core library standards.) The dataset package and the dataset class add such metadata with the help of interface functions to the base R attributes part of the data.frame (or tibble, data.table) object. This is a useful feature that may go through code or interface refinement, but works well and should not significantly change. The current version of dataset is too specific because it tries to map the SDMX definitions of the data structure, which goes far beyond the needs of a well-described tidy dataset. The datacube statistical vocabulary and standard provides further structural information about how we can subset or slice meaningfully a statistical dataset. While such structural information is indispensable for certain statistical datasets, it may be completely useless when it contains statistically not aggregated data ("microdata"), or collections data as a primary statistical data source. To demonstrate the long-term ambition, we want to develop the following functionality: - The `wbdataset` package for a correct format for collecting data from Wikibase, a semantic database widely used to share micro-data. This package can later be generalised to work with further data models that have a similar aim. - The future `datacube` package should support the semantic needs of datasets that contain statistically processed information from a dataset that was collected via the `wbdataset` function. The semantic needs are different, because the observational units will change after statistical aggregation. - The dataset package should provide a general framework for handling dataset-level microdata and the basic semantic needs of identifying rows and columns. ### The wbdataset package concept The [Wikibase Data Model](https://www.mediawiki.org/wiki/Wikibase/DataModel) is a relatively simple and flexible data model. It works with concepts and properties as relationships among concepts. A tidy dataset that applies the _Wikibase Data Model_ can be described (using the definitions of Wikidata, the world's largest public database created with Wikibase) the following: The key column defines the data subject or statistical unit with a QID identifier. This identifier is denoted with a capital Q followed by an integer number. The QID is unique in one instance of a Wikibase database. The full identifier contains the URL of this database and the QID. For example, unambiguously defines the small country of Andorra in many natural languages, and with the connection of various actionable, persistent identifiers. (Check out on the linked page for example: _sovereign microstate between France and Spain, in Western Europe_, or or `https://isni.org/isni/000000012150090X/`. ```{r installwbdataset, eval=FALSE} # install.packages("devtools") devtools::install_github("dataobservatory-eu/wbdataset") ``` ```{r examplewbdataset, eval=FALSE} library(wbdataset) get_item( qid = c("Q228", "Q347"), language = c("en", "nl"), creator = person("Jane Doe"), title = "Small Countries" ) ``` With a `wbdataset` object, it is important that the key column is a QID. Following the notations of tidyverse, instead of the tibble::rowid_to_colum, we create a`wbdataset::qid_to_column function that creates an identifier for each row. The variables or columns bring the observational unit (data subject or statistical subject) into a pre-defined relationship with the cell value. These pre-defined relationships are identified in the Wikibase Data Model with a property or PID identifier. The PID is denoted with a capital P followed by an integer number. The PID is unique in one Wikibase-created database. For example, or simply [P569](https://www.wikidata.org/wiki/Property:P569) refers to the date of birth, and [P276](https://www.wikidata.org/wiki/Property:P276) denotes a location (of the data subject), and [P3629](https://www.wikidata.org/wiki/Property:P3629) the age of subject at event. ```{r smallcountrydataset2} small_country_musicians <- data.frame( qid = c("Q275912", "Q116196078"), label = c("Marta Roure", "wavvyboi"), P276 = c("Andorra", "Lichtenstein"), P569 = c(as.Date("1981-01-16"), as.Date("1998-04-28")) ) ## And the age small_country_musicians$P3629 <- 2024 - as.integer( substr(small_country_musicians$P569, 1, 4) ) small_country_musicians ``` Recalling the tidy data definition, the cell values may be numbers or "strings for qualitative information". We extend the possibilities to further options following the RDF standards: - Numbers that do not require further semantic definition and interpretation - Strings that do not offer further semantic definition (though they may need it!) - Time concepts that follow the ISO time definitions - URIs of any concepts that contain their definition (extending the possibility of using strings that refer to well-defined things.) Therefore, the column types should have an unambiguous mapping to RDF: - Numeric, integer and real types are not interpreted but must be serialised to xs:integer or xs:decimal. - Logical should not be interpreted but serialised as xs::boolean. - Various date and time classes must be serialised to xs:date or xs:time. - Well-defined concepts should contain a URI, or a curie (a shortened version of the URI) should be serialised as xs:anyURI. ```{r smallcountrydataset3} small_country_musicians <- data.frame( qid = c("Q275912", "Q116196078"), label = c("Marta Roure", "wavvyboi"), P276 = c("Q228", "Q347"), P569 = c(as.Date("1981-01-16"), as.Date("1998-04-28")) ) ## And the age small_country_musicians$P3629 <- 2024 - as.integer( substr(small_country_musicians$P569, 1, 4) ) small_country_musicians$P569 <- dataset::xsd_convert(small_country_musicians$P569) small_country_musicians$P3629 <- dataset::xsd_convert(small_country_musicians$P3629) small_country_musicians ``` We can also add the prefix to any concepts that are well-defined: ```{r definecountry} small_country_musicians$P276 <- paste0("https://www.wikidata.org/wiki/", small_country_musicians$P276) small_country_musicians[, 2:4] ``` ### The future datacube package The current `dataset` class and functionality should relax the modelling of the DataStructureDefinition of SDMX because it makes only sense for statistically aggregated data. Currently, the dataset package does not support the concept of a slice, and it is a bit awkward to support multi-dimensional datacubes. Such functionality should be moved to a low-priority new package. ### What should remain in the dataset package? The specifications of the `wbdataset` package should all be placed into the dataset package whenever it is not specific to the Wikibase Data Model. They should be co-developed with `wbdataset` to provide a well-defined interface towards a global data system (Wikidata and its "private clones" of Wikibase instances.) The `wbdataset` package should allow a simple, natural way to import data from Wikidata or a Wikibase instance, and it should also provide a simple interface to send data back to such a semantic database with ease. The distinction between wb-dataset and dataset is justified because a stripped-down dataset package can still work well in many SDMX or other contexts, albeit without the full functionality of supporting statistical slicing or API support to a specific SDMX-compatible web service. An R package that allows the creation of semantically rich SDMX-compatible datasets with only manual downloading or uploading functionality would still be a great improvement in implementing open science interoperability and reusability for such datasets.