Data

Article Corpus and Data Model

Author

Affiliation

Moritz Twente

Universität Basel

Modified

January 7, 2026

This page contains a description of the data used in this project, aiming to provide information on both the workflow and the content produced by running it, as well as to present the data model for the resulting corpus and other data output.

Article Corpus

The corpus is created based on the bibliography provided by Ruedin and Hanak (2008, 216–20). I did not include all publications into the data set but focused on the articles published in newspapers and professional journals. This focus leaves out ‹gray literature› – mainly technical documents, planning reports, and printed conference proceedings – and concentrates on texts that were intended for a wider audience and can be understood as part of relatively regular publishing activity.

The 189 articles by Hans Marti that were considered¹ for creating the corpus are listed in Table 1. All of them are accessible via e-periodica.ch or e-newspaperarchives.ch.

Table 1

Workflow

A multi-step workflow (Figure 1) is used to scrape the articles from the web. After some cleaning, the quality of the text files is good enough to be used for modelling topics with quanteda (Benoit et al. 2018). Topic models and the result of the analysis are summarised in a report which is rendered with Quarto and deployed to GitHub Pages. The repository as a whole is archived on Zenodo.

%%{ init: {  'themeVariables': { 'edgeLabelBackground': '#fcfcfc'} } }%%
flowchart TD

    subgraph Resources
        RN2[("e-periodica.ch<br>e-newspaperarchives.ch")] -->|Article IDs| RN13
        RN1[(Literature Research)] -->|Bibliography| RN13
        RN1 -->|Biographical Context| RN13
    end

    CB1 -->|Report| RN9
    RN14 -->|&ast;.txt<br>Files| CB2

    subgraph Data Storage

        RN9(["modelling-marti <br>(GitHub Repository)"])
        RN9 -->|Quarto| RN11([Online<br>Documentation])
        RN9 --> RN10([Zenodo])
        RN5([Zotero Bibliography])

    end    

    subgraph Analysis
        RN14 -->|Processed<br>Article Texts| CB1[Analysis<br>&lpar;Quanteda&rpar;]
        CB2([Corpus]) --> RN9

        RN2 -.->|Download| RN14

    end

    RN13([docs/articles_metadata.csv]) --> RN9
    RN1 ------> RN5
    RN13 --> RN14([Full Articles])

style CB1 fill:#90C987
style CB2 fill:#7BAFDE
style RN1 fill:#D9CCE3
style RN2 fill:#D9CCE3
style RN5 fill:#F6C141
style RN9 fill:#F7F056
style RN10 fill:#F7F056
style RN11 fill:#F6C141
style RN13 fill:#7BAFDE
style RN14 fill:#7BAFDE

Figure 1: Workflow used for assembling and analysing the dataset.

Data Model

Article Data

All data is available in csv format. For this file format, metadata is provided in json files according to the W3C Metadata Vocabulary for Tabular Data (W3C 2022). In the R workflow, the metadata object is used to create a list of articles which includes an URL to download the textual data into another R object text_data. The resulting marti_corpus contains the full articles, metadata on those articles and information on Marti’s work status at the time of publication for each article (Figure 2).

classDiagram
    direction LR

    metadata <|-- articles
    articles <.. marti_corpus
    text_data <.. marti_corpus
    raw_text "1..*" o-- text_data
    berufslaufbahn <|-- marti_corpus

    class metadata {
        id : str
        title : str
        publication : str
        date : str &lpar;%d.%m.%y&rpar;
        language: str &lpar;ISO 639-1&rpar;
        standalone: bool
        first_row: int
        last_row: int
        archive_id: str
        note: str
    }

    class articles {
        id : str
        title : str
        publication : str
        date : str &lpar;%d.%m.%y&rpar;
        language: str &lpar;ISO 639-1&rpar;
        standalone: bool
        first_row: int
        last_row: int
        archive_id: str
        note: str
        url: str
    }

    class text_data {
        doc_id: str
        text: str
    }

    class raw_text {
      &ast;&period;txt &lpar;article texts&rpar;
    }

    class berufslaufbahn {
        Beruf : str
        Start : str &lpar;%Y-%m-%d&rpar;
        Ende : str &lpar;%Y-%m-%d&rpar;
    }

    class marti_corpus {
        doc_id: str
        text: str
        title: str
        publication: str
        date : str &lpar;%Y-%m-%d&rpar;
        language: str &lpar;ISO 639-1&rpar;
        VLP: bool
        SBZ: bool
        Delegierter: bool
        Gemeinderat: bool
        Pensionierung: bool
        fachpublikum: bool
        pol_mandat: bool
    }

style marti_corpus fill:#F7CB45
style text_data fill:#FFEAAE
style articles fill:#D1BBD7

Figure 2: Chart illustrating variables and metadata included in the corpus used in this project.

Corpus Metadata

The marti_corpus corpus object is stored as csv file in this repository. A json metadata file is available with accompanying information. The content of the corpus is described below.

Description of variables in `marti_corpus`
variable	datatype	description	example value
`doc_id`	string	unique identifier for each article	`NZZ_19550205_0010`
`text`	string	full processed text body as plain text	Auf die in der umstrittenen Basler…
`title`	string	title of the article as published	Der Traum von der neuen Stadt
`publication`	string	name of the journal/newspaper	NZZ
`date`	string	date of publication, `%Y-%m-%d` format	`1955-02-05`
`language`	string	main language the article is written in, ISO 639-1 code	`de`
`VLP`	boolean	Marti worked at the VLP at the time of publication	`FALSE`
`SBZ`	boolean	Marti worked at the Bauzeitung at the time of publication	`TRUE`
`Delegierter`	boolean	Marti worked as Delegate of the City Government at the time of publication	`FALSE`
`Gemeinderat`	boolean	Marti was member of the City Parliament at the time of publication	`FALSE`
`Pensionierung`	boolean	Marti was retired at the time of publication	`FALSE`
`fachpublikum`	boolean	architecture/planning professionals are the main audience of the publication	`FALSE`
`pol_mandat`	boolean	Marti held a public office at the time of publication	`FALSE`

All information is the result of the processing workflow and taken from one of the data objects in earlier steps of the R workflow which are listed in more detail in the following sections.

Article Metadata

`metadata`

The metadata object in the R workflow is a data frame containing the articles’ metadata, mostly taken from the bibliography provided in Ruedin and Hanak (2008, 216–20), enhanced by extensive research on Marti and his work. The metadata object is stored as csv file in this repository.

Description of variables in `metadata`
variable	datatype	description	example value
`id`	string	unique identifier for each article	`NZZ_19550205_0010`
`title`	string	title of the article as published	Der Traum von der neuen Stadt
`publication`	string	name of the journal/newspaper	NZZ
`date`	string	date of publication as taken from scanned documents, `%d.%m.%y` format	`05.02.55`
`language`	string	main language the article is written in, ISO 639-1 code	`de`
`standalone`	boolean	complete article in one item (as opposed to publications spread over multiple issues)	`TRUE`
`first_row`	integer	number of first row in *.txt file containing actual content (i.e., not a headline)	3
`last_row`	integer	number of last row in *.txt file containing actual content (i.e., end of article)	27
`archive_id`	string	identifier used in permalinks at e-newspaperarchives.ch	`NZZ19550205-01.2.30.2`
`note`	string	notes regarding the transcription or the article itself	Bildbeschreibung in den Text gerutscht

`articles`

The articles object in the R workflow is a data frame based on the metadata basis data frame, but additionally contains an URL pointing to the full texts of each article on either e-periodica.ch or e-newspaperarchives.ch.

Description of variables in `articles`
variable	datatype	description	example value
`id`	string	unique identifier for each article	`NZZ_19550205_0010`
`title`	string	title of the article as published	Der Traum von der neuen Stadt
`publication`	string	name of the journal/newspaper	NZZ
`date`	string	date of publication as taken from scanned documents, `%d.%m.%y` format	`05.02.55`
`language`	string	main language the article is written in, ISO 639-1 code	`de`
`standalone`	boolean	complete article in one item (as opposed to publications over multiple issues)	`TRUE`
`first_row`	integer	number of first row in *.txt file containing actual content (i.e., not a headline)	3
`last_row`	integer	number of last row in *.txt file containing actual content (i.e., end of article)	27
`archive_id`	string	identifier used in permalinks at e-newspaperarchives.ch	`NZZ19550205-01.2.30.2`
`note`	string	notes regarding the transcription or the article itself	Bildbeschreibung in den Text gerutscht
`url`	string	URL to access the full article (raw text or PDF)	`https://...`

Article Text

`text_data`

The text_data object in the R workflow is a data frame containing all the articles’ full texts after retrieval from either e-periodica.ch or e-newspaperarchives.ch and subsequent semi-automatic cleaning.

Description of variables in `text_data`
variable	datatype	description	example value
`doc_id`	string	unique identifier for each article	`jub-002_1964_26__231_d`
`text`	string	full processed text body as plain text	Im Jahre 1947 wurde in Baden…

Professional Career

Selected events on a timeline of Hans Marti’s professional career are stored in the corpus and used as corariate variables for topic modelling. The dates are stored in a csv file accompanied by a json metadata file. The data is based on information by Lendi (2018), Böcker (2007) and Koll-Schretzenmayr (2008).

Description of variables in `berufslaufbahn`
variable	datatype	description	example value
`Beruf`	string	unique identifier for each job	VLP
`Start`	string	start date of job, `%Y-%m-%d` format	`1945-01-01`
`Ende`	string	end date of job, `%Y-%m-%d` format	`1948-01-01`

Geodata

For visualisation purposes, this project uses geodata to create a project map as part of the analysis. The map shows each project that Marti and/or his office were involved in. Information on the projects is taken from Ruedin and Hanak (2008, 222–24) and stored in data/geodata/marti-geodata-json.geojson. For printing a static map to the PDF version of the analysis, I use a topographic map of Switzerland provided by swisstopo (2025) which is clipped using boundary data from OpenStreetMap (data/geodata/CH_outline.geojson). See Figure 3 for an overview of the structure of the geojson files.

classDiagram
    direction TD

    class marti-geodata-json {
        id: int
        project: str
        period: str
        location: str
        coordinates: str &lpar;EPSG:2056/CH1903+&rpar;
    }

    class CH_outline {
        type: str
        coordinates: str &lpar;CRS84&rpar;
    }

Figure 3: Chart illustrating variables and metadata included in the geodata used in this project.

Planning Project Inventory

The list of projects by Marti and his office is available as geojson file. The coordinates assigned to each project are not corresponding to any actual planning activities but point to a generic point within each project perimeter. The values for the other variables are taken from the project inventory by Ruedin and Hanak (2008, 222–24). The dataset is using Swiss Landeskoordinaten (EPSG:2056, CH1903+/LV95). A QGIS metadata file is available on GitHub alongside the geojson file itself.

Description of variables in `marti-geodata-json`
variable	datatype	description	example value
`id`	integer	unique identifier for each project	`104`
`project`	string	project title/description (in German)	Ortsplanung, Verkehrsplanung, Beratung Nationalstrassenführung, Quartierpläne
`period`	string	timeframe of project duration	1960–1975
`location`	string	name of location, mostly municipalities	Murten
`coordinates`	string	E/N coordinates of point geometry, `EPSG:2056` (Swiss CH1903+/LV95 format)	`2575516.085672613698989, 1197518.942096951650456`

Map of Switzerland

A map of Switzerland used in the PDF version of the report is produced using geodata from OpenStreetMap (OpenStreetMap Contributors 2025), more specifically: Relation #51701 v545. This data (Swiss border geometry) is available in a geojson file. The dataset is using WGS 84 coordinates (OGC:CRS84).

Description of variables in `CH_outline`
variable	datatype	description	example value
`type`	integer	feature geometry type	`MultiPolygon`
`coordinates`	string	E/N coordinates of point geometries, `OGC:CRS84` (WGS 84 format)	`9.1572929, 47.6659088`

References

Benoit, Kenneth, Kohei Watanabe, Haiyan Wang, Paul Nulty, Adam Obeng, Stefan Müller, and Akitaka Matsuo. 2018. “Quanteda: An R Package for the Quantitative Analysis of Textual Data.” Journal of Open Source Software 3 (30): 774. https://doi.org/10.21105/joss.00774.

Böcker, Dagmar. 2007. “Marti, Hans.” Historisches Lexikon Der Schweiz (HLS). https://hls-dhs-dss.ch/de/articles/027384/2007-10-22/.

Koll-Schretzenmayr, Martina. 2008. “Hans Marti Und Die Jugendjahre Der Schweizerischen Landesplanung – Eine Zeitreise.” In Hans Marti – Pionier Der Raumplanung, edited by Claude Ruedin and Michael Hanak, 32–37. Zürich: gta Verlag.

Lendi, Martin. 2018. Geschichte Und Perspektiven Der Schweizerischen Raumplanung: Raumplanung Als Öffentliche Aufgabe Und Wissenschaftliche Herausforderung. Zürich: vdf Hochschulverlag AG an der ETH Zürich.

OpenStreetMap Contributors. 2025. “OpenStreetMap.” Geospatial Database. https://www.openstreetmap.org.

Ruedin, Claude, and Michael Hanak, eds. 2008. Hans Marti – Pionier Der Raumplanung. Zürich: gta Verlag.

swisstopo. 2025. “swissALTI3d: Das Hoch Aufgelöste Terrainmodell Der Schweiz.” Bern: Bundesamt für Landestopographie. https://www.swisstopo.admin.ch/de/hoehenmodell-swissalti3d.

W3C. 2022. “Model for Tabular Data and Metadata on the Web.” https://w3c.github.io/csvw/syntax/.

Footnotes

All non-German articles were removed before executing the topic modelling workflow.↩︎