Periodic Reporting for period 1 - REWIND (Rereading European Cultural Heritage in Latin American Women Writers’ Travel Literature of the early 20th century: contrasting testimonies to build inclusive historical discourses)
Période du rapport: 2023-01-01 au 2024-12-31
Résumé du contexte et des objectifs généraux du projet
The REWIND project is an international initiative that employs an innovative, interdisciplinary methodology rooted in Digital Humanities. It uses travel literature as a historical source to explore European cultural heritage through decolonial and gendered perspectives. The project's workflow integrates XML-TEI encoding, GLAM-Linked Open Data, and Geographic Information Systems to develop an ontological model connecting early 20th century European cultural heritage with the semantic web. Additionally, aspects of the reception of dissonant heritage were analysed using techniques such as sentiment analysis, collocation studies, and semantic field mapping, among others from Corpus Linguistics. The project's aim was to amplify the voices of Latin American women engaged in feminism and sociocultural diversity, positioning them as agents in constructing historical narratives about European cultural heritage. This approach confronted the authorised heritage discourse, promoting decolonial and inclusive historical narratives. By exploring diverse historical discourses, the project adopted a polyphonic approach to the reception of heritage. It went beyond simply adding women's testimonies or listing cultural productions they created, aiming to construct a more plural history. However, it faced challenges in countering the authorised heritage discourse of the period. Identifying the authors as Latin American women does not automatically exclude perspectives influenced by whiteness, androcentrism, patriarchy, or Eurocentrism, which may reflect their 'European' education. This complexity was further compounded by the elitism present in certain social circles, which constituted the primary Latin American demographic able to undertake European leisure travel during the first decades of the 1900s.
The REWIND project structured into specific research and innovation objectives [R&IO]:
R&IO1: To reconstruct the travellers' itineraries [WP2]. Geo-semantic annotation of travel books by identifying and mapping the European cultural heritage elements mentioned by authors.
The XML markup language and the standards proposed by the Text Encoding Initiative (TEI) were used to annotate geo-semantically the books in the REWIND project corpus. To ensure interconnected information, each cultural heritage element was assigned a unique identifier using the @xml:id attribute, linked to an index with detailed information about each item through the @ref attribute. This information included geographical coordinates, typology, URI, and other relevant data. The encoding procedure was thoroughly documented in the file REWIND_Encoding_Guidelines.pdf and is available in the project's Zenodo repository.
R&IO2: To connect the information using Linked Open Data (LOD) [WP2]. An ontology based on CIDOC-Conceptual Reference Model.
The REWIND_Ontology.owx was designed using the Web Ontology Language (OWL), a markup language for publishing and sharing data through ontologies on the web. The OWL language aims to facilitate a markup model built on RDF and encoded in XML. The model CIDOC-CRM was employed to ensure data interoperability, providing an extensible ontology for concepts and information in cultural heritage and museum documentation, as travel books often reflected a traditional conception of heritage. Additionally, the controlled vocabulary from the ROSSIO Infrastructure was used to create typology entities corresponding to the E55 Type class, the GeoNames gazetteer for entities in the E53 Place class, the VIAF authority file for entities in the E21 Person class, and Wikidata for all identified cultural heritage elements mentioned in the studied travel books, associated with E22 Human-Made Object and E28 Conceptual Object classes. The XML encoding was based on the ontological model, so exporting the information from the books in a machine-readable format to the ontological model was straightforward. This approach enabled the study of cultural heritage items through patterns and relationships that are difficult for humans to detect. The mapping procedure was thoroughly documented in the file REWIND_Ontology_Guidelines.pdf and is accessible in the project's Zenodo repository.
R&IO3: To recognise impressions associated with the European cultural heritage [WP3]. Sentiment analysis is applied to explore the evocative vocabulary of the cultural heritage descriptions.
A sentiment analysis was conducted on all cultural heritage elements potentially associated with the concept of dissonant heritage, such as the Columbus monuments in Barcelona, Genoa, Seville, and Madrid, due to their connection to the colonisation of the Americas, as well as other elements related to political issues, such as the French Revolution. In this context, the results related to the Monastery of El Escorial and the statue of Isabella II of Spain were particularly interesting. However, it was identified that this type of approach presented biases. The NRC Emotion Lexicon (Mohammad & Turney, 2013) was specifically chosen because it had been manually annotated. Still, for example, when analysing the Mosque of Córdoba, the emotional tone was consistently negative, despite the text praising the historic building. This issue led to adjustments in the analysis, such as studying the semantic fields of words associated with the eight 'basic emotions' defined by Plutchik (2001) and the lexical placement of adjectives. Ultimately, it was decided not to share the scripts and workflow on Zenodo to prevent the identified biases from being perpetuated in future studies.
R&IO4: To determine the relationship between European cultural heritage and otherness processes [WP3]. Gender decolonial theories emphasise the need to make non-European women visible as cultural agents in historical narratives to break the Eurocentric, patriarchal, and androcentric cultural canon.
There is no single historical narrative; however, a predominant one often emerges depending on the political context. The REWIND project analysed the general historiographical trends documented in the most impactful publications on cultural heritage between 1900 and 1930, alongside information from popular travel guides by Baedeker (1898), Bradshaw (1894), and Murray (1890). Once the highlighted monuments or works of art were identified within authorised heritage discourse, along with the historical events or processes with which they were associated, it became possible to compare this data with references in the studied travel books to explore discrepancies between official and hegemonic historical narratives and the reception of Latin American women writers.
The REWIND project structured into specific research and innovation objectives [R&IO]:
R&IO1: To reconstruct the travellers' itineraries [WP2]. Geo-semantic annotation of travel books by identifying and mapping the European cultural heritage elements mentioned by authors.
The XML markup language and the standards proposed by the Text Encoding Initiative (TEI) were used to annotate geo-semantically the books in the REWIND project corpus. To ensure interconnected information, each cultural heritage element was assigned a unique identifier using the @xml:id attribute, linked to an index with detailed information about each item through the @ref attribute. This information included geographical coordinates, typology, URI, and other relevant data. The encoding procedure was thoroughly documented in the file REWIND_Encoding_Guidelines.pdf and is available in the project's Zenodo repository.
R&IO2: To connect the information using Linked Open Data (LOD) [WP2]. An ontology based on CIDOC-Conceptual Reference Model.
The REWIND_Ontology.owx was designed using the Web Ontology Language (OWL), a markup language for publishing and sharing data through ontologies on the web. The OWL language aims to facilitate a markup model built on RDF and encoded in XML. The model CIDOC-CRM was employed to ensure data interoperability, providing an extensible ontology for concepts and information in cultural heritage and museum documentation, as travel books often reflected a traditional conception of heritage. Additionally, the controlled vocabulary from the ROSSIO Infrastructure was used to create typology entities corresponding to the E55 Type class, the GeoNames gazetteer for entities in the E53 Place class, the VIAF authority file for entities in the E21 Person class, and Wikidata for all identified cultural heritage elements mentioned in the studied travel books, associated with E22 Human-Made Object and E28 Conceptual Object classes. The XML encoding was based on the ontological model, so exporting the information from the books in a machine-readable format to the ontological model was straightforward. This approach enabled the study of cultural heritage items through patterns and relationships that are difficult for humans to detect. The mapping procedure was thoroughly documented in the file REWIND_Ontology_Guidelines.pdf and is accessible in the project's Zenodo repository.
R&IO3: To recognise impressions associated with the European cultural heritage [WP3]. Sentiment analysis is applied to explore the evocative vocabulary of the cultural heritage descriptions.
A sentiment analysis was conducted on all cultural heritage elements potentially associated with the concept of dissonant heritage, such as the Columbus monuments in Barcelona, Genoa, Seville, and Madrid, due to their connection to the colonisation of the Americas, as well as other elements related to political issues, such as the French Revolution. In this context, the results related to the Monastery of El Escorial and the statue of Isabella II of Spain were particularly interesting. However, it was identified that this type of approach presented biases. The NRC Emotion Lexicon (Mohammad & Turney, 2013) was specifically chosen because it had been manually annotated. Still, for example, when analysing the Mosque of Córdoba, the emotional tone was consistently negative, despite the text praising the historic building. This issue led to adjustments in the analysis, such as studying the semantic fields of words associated with the eight 'basic emotions' defined by Plutchik (2001) and the lexical placement of adjectives. Ultimately, it was decided not to share the scripts and workflow on Zenodo to prevent the identified biases from being perpetuated in future studies.
R&IO4: To determine the relationship between European cultural heritage and otherness processes [WP3]. Gender decolonial theories emphasise the need to make non-European women visible as cultural agents in historical narratives to break the Eurocentric, patriarchal, and androcentric cultural canon.
There is no single historical narrative; however, a predominant one often emerges depending on the political context. The REWIND project analysed the general historiographical trends documented in the most impactful publications on cultural heritage between 1900 and 1930, alongside information from popular travel guides by Baedeker (1898), Bradshaw (1894), and Murray (1890). Once the highlighted monuments or works of art were identified within authorised heritage discourse, along with the historical events or processes with which they were associated, it became possible to compare this data with references in the studied travel books to explore discrepancies between official and hegemonic historical narratives and the reception of Latin American women writers.
Travail effectué depuis le début du projet jusqu’à la fin de la période considérée dans le rapport et principaux résultats atteints jusqu’à présent
WP1 Corpus and dataset
The configuration of the REWIND corpus was particularly challenging. The selection of sources was based on several criteria to ensure the research aligned coherently with the framework of decolonial gender theory. Thus, all corpus texts were authored by Latin American women, published in Spanish in book format, and written during their travels to predominantly the same European countries in the 20th century, prior to the Second World War. These authors identified with the feminist cause and advocated for cultural diversity. In this way, Inés Echeverría (1937), Clorinda Matto (1909), Zoila Aurora Cáceres (1911), María Enriqueta Camarillo (1930), and Isabel Carrasquilla (2011), become sociocultural interpreters of European cultural heritage, building bridges between their cultural origin and Europe.
The REWIND dataset aligns with the concept of 'deep data' (Štular & Belak, 2022), as the geo-semantic analysis required a close engagement with the text, with each annotation necessitating interpretation. The workflow for encoding texts in XML-TEI format began with the digitisation of paper travel books to generate PDF files. All texts were sourced from public digital repositories, except for Oasis de Arte, which was requested by the National Library of Spain to be digitised. Subsequently, Optical Character Recognition (OCR) techniques were employed to convert the images into plain text files in TXT format, which then served as the basis for encoding.
Before encoding the texts in XML-TEI, a conceptual analysis was conducted to develop the ontological model based on the CIDOC-CRM proposal. The semantic labelling used was grounded in the classes defined in the ontology, ensuring that the information was correlated and interoperable. The UNESCO typology was applied to catalogue different cultural heritage items, distinguishing between intangible cultural heritage (E28 Conceptual Object) and movable (E22 Human-Made Object) and immovable (E53 Place) material cultural heritage. Additionally, the ROSSIO Thesaurus was used to categorise the different elements of European cultural heritage. This classification aligned with controlled vocabulary, enhancing integration and enabling the use of Linked Open Data (LOD). Controlled vocabularies were also employed to limit the meaning of terms used to define an object, addressing natural language issues such as synonymy or polysemy.
The REWIND dataset was created by encoding travel books using XML markup language and the standards proposed by the Text Encoding Initiative. On the one hand, a modernised diplomatic digital edition of each publication was produced, ensuring that not only the structure of the books was annotated, but also that the original formatting of the text (e.g. italics, capital letters, or indentation) and any typographical errors were preserved. Additionally, the Spanish language was also adapted to the most recent edition of the Ortografía de la Lengua Española (RAE, 2010), as the sentiment analysis relied on a contemporary Spanish Emotion Lexicon. On the other hand, tagging was used to identify people, places, objects, and events, turning each book into a database that could be queried using XQuery expressions. This annotation system enabled differentiation of all cultural heritage elements by linking them, whenever possible, with geographical coordinates through the GeoNames gazetteer and to LOD from VIAF and Wikidata. Creating semantically well-structured data from diverse sources was a complex task. To address this, all information was centralised in an index that referenced the same object across different travel books by means of a unique identifier, implemented through the @xml:id attribute.
For elements of material cultural heritage, the tag was used, corresponding to the E22 Human-Made Object class of the ontological model. The @type attribute specified different instances of the E55 Type class based on the ROSSIO Thesaurus, enabling precise categorisation of cultural heritage elements. This approach ensured semantic clarity and supported interoperability within the ontological framework. For example, instances of E55 Type could differentiate between categories such as ‘painting’, 'sculpture’, or 'book’, providing a structured and standardised way to describe diverse heritage elements. Immovable heritage items were considered places, as they could contain movable heritage, and were tagged with . The @type attribute specified whether the place was a geopolitical entity (e.g. city, country, or region) or a cultural heritage element (e.g. archaeological site, historical building, or museum). The tag was used to designate geographical features, such as mountains, lakes, capes, or rivers. To encode the documents, the Visual Studio Code editor was used. This code editor is free and open-source software, although the official download is proprietary software released under the Microsoft Software License.
Outcome and results: Comino Comino, A. (2024). REWIND Corpus (1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.14093621(s’ouvre dans une nouvelle fenêtre)
WP2 Geo-semantic annotation of travel books
Historical narrative sources are often characterised by numerous interconnected references to persons, places, institutions, events, dates, etc. For this reason, a knowledge graph based on the CIDOC-CRM (v. 7.1.3 February 2023), a model developed by ICOM and converted into the ISO 21127:2023 standard, was chosen to structure historical information in a manner compatible with the semantic web, rather than using a relational database. The CIDOC-CRM consists of 81 classes and 160 properties, structured hierarchically to represent concepts and information related to cultural heritage and museum documentation. Once the text was encoded in a machine-readable format, it became possible to extract data relating to the elements of European cultural heritage mentioned in the travel books and integrate them into the ontological model. The REWIND project focused on event instances when modelling cultural heritage data contained in travel books, with attention given to what is observed, when, where, and by whom.
An ontology provides a model for the management and semantic representation of knowledge, defining a structured set of concepts and describing the relationships between them through classes and properties, to which values are associated. While humans cannot easily detect interconnections between places, objects, events, and people, digital approaches enable quantification and pattern recognition. Additionally, an ontology allows for linking the historical content intrinsic to the text with external information through the use of LOD. In this context, CIDOC-CRM was used, as the heritage conception in the analysed travel books had a museographic character.
In the REWIND project, a workflow was implemented consisting of an initial phase to identify elements of European cultural heritage mentioned in travel books through XML-TEI encoding, followed by a second phase in which these data were exported to the ontological model for analysis. Once the text was encoded in a machine-readable format using XML-TEI, information related to the elements of European cultural heritage referenced by the authors in their travel books was extracted and integrated into the ontological model. To access the XML documents, the open-source software eXist-db was used. The XQuery language facilitated the browsing of XML-TEI files to locate all heritage references and transfer them to a knowledge graph created with Protégé, a free open-source ontology editor and framework for building intelligent systems developed by Stanford University. This process was relatively straightforward because the semantic labelling employed by XML-TEI was based on the classes and properties defined in the ontology, ensuring that the information was both correlated and interoperable.
Outcome and results: Comino Comino, A. (2024). REWIND Ontology. Zenodo. https://doi.org/10.5281/zenodo.14577943(s’ouvre dans une nouvelle fenêtre)
WP3 Opinion mining and comparative analysis
In a memory space, diverse narratives coexist and compete to shape how the past is represented or recounted, frequently giving rise to contradictions, conflicts, and tensions. Several elements aligned with the concept of dissonant heritage, such as monuments associated with the European colonisation of the Americas, were selected to compare the authors' narratives with authorised heritage discourse. Specifically, in the case of Spain, texts by intellectuals such as Rafael Altamira (1900-1911) and Emilio Castelar (1892), alongside references from travel guides by Baedeker (1898), Bradshaw (1894), and Murray (1890), were analysed to explore discrepancies between official and hegemonic historical narratives and the reception of colonial monuments by Latin American women writers. The aim was to carry out a reverse reading; that is, to examine whether the authorised heritage discourse present in history books and reflected in monuments was also transmitted to travel books through the reception process by travellers. Furthermore, the objective was to assess the influence of the travel guides consulted by the authors on their interpretations, considering that the REWIND project sought alternative testimonies for constructing historical knowledge about cultural heritage.
On the one hand, to contrast discourses from the perspective of otherness, keywords such as 'our', 'hero', 'fatherland', 'distinguished man', and 'conquest' were selected, and their semantic contexts were analysed. On the other hand, sentiment analysis was applied to descriptions of dissonant heritage elements, as emotional responses to heritage during travel have been associated with psychosomatic reactions, including Stendhal, Jerusalem, and Kinkaku-ji syndromes. Starting from the premise that most cultural objects are designed to provoke an aesthetic response, travel books provide valuable resources for sentiment analysis. However, in literary texts, emotions function as an expressive device aimed at eliciting a response from the reader. Therefore, when analysing travel narratives, it was necessary to distinguish between the affective dimension associated with autobiographical narrative and the sensations evoked during reading. Nonetheless, the study of emotions proves highly complex, as their expression is contingent upon the historical context. It was thus essential to consider the psychological theories and sociocultural environments of the period. Using this approach, the REWIND project addressed questions such as: Which heritage elements were selected in travel books? What power relations did they perpetuate? What versions of memory, identity, or sense of place did they legitimise?
Analysing documents in Spanish posed a challenge, as most digital sentiment analysis applications and methods had been developed for English texts. Methodologies, workflows or techniques needed to be adapted. The project adopted a proposal that utilised the R programming language, the Tidytext and Syuzhet packages, and the NRC Emotion Lexicon (Isasi, 2021) to detect and evaluate the emotions present in Spanish-language travel books. The NRC Emotion Lexicon (Mohammad & Turney, 2013) contains 14,182 unigrams manually annotated with both positive or negative sentiment valences and the emotions of anger, anticipation, disgust, fear, joy, sadness, surprise, and trust. Although this procedure was a straightforward approach to the study of the eight 'basic emotions' defined by Plutchik (2001), it allowed for the observation of a connection between heritage and emotionality.
Outcome and results: A set of scripts in R was developed to analyse the texts, but they were ultimately not shared in the project's Zenodo repository due to the identification of biases that were not intended to be perpetuated. Sentiment analysis applied to texts in Spanish proved highly complex, particularly as the REWIND project focused on early 20th century Spanish-language texts from various Spanish-speaking countries. While the expression of emotions is inherently personal, it is also shaped by cultural codes linked to social status and the evolution of language. These factors posed significant challenges when attempting to model emotions associated with descriptive texts about cultural heritage elements. To address these biases, a lexical collocation analysis was conducted to explore associations between adjectives with positive or negative valences and cultural heritage objects or historical figures. Furthermore, the analysis accounted for context by incorporating the effects of negation adverbs.
Bibliography:
Altamira, R. (1900–1911). Historia de España y de la civilización española. Juan Gili.
Baedeker, K. (1898). Spain and Portugal. Karl Baedeker.
Cáceres, A. (1911). Oasis de arte. Garnier Hermanos.
Camarillo, M. E. (1930). Brujas, Lisboa, Madrid. Espasa Calpe.
Carrasquilla, I. (2011). Impresiones de viaje escritas por una abuela para sus nietos. Universidad EAFIT.
Castelar, E. (1892). Historia del descubrimiento de América. [s.n.].
Charnock, R. S. (1894). Bradshaw's illustrated hand-book to Spain and Portugal: A complete guide for travellers in the Peninsula. W. J. Adams & Sons.
Echeverría, I. (1937). Entre dos siglos. Ediciones Ercilla.
Ford, R. (1890). Handbook for travellers in Spain (7th ed.). J. Murray.
Isasi, J. (2021). Análisis de sentimientos en R con ‘syuzhet’. Programming Historian en español, 5. https://doi.org/10.46430/phes0051(s’ouvre dans une nouvelle fenêtre)
Matto, C. (1909). Viaje de recreo: España, Francia, Inglaterra, Italia, Suiza, Alemania. F. Sempere.
Mohammad, S. & Turney, P. D. (2013). Crowdsourcing a Word–Emotion Association Lexicon. Computational Intelligence, 29(3), 436–465. https://doi.org/10.1111/j.1467-8640.2012.00460.x(s’ouvre dans une nouvelle fenêtre)
Plutchik, R. (2001). The nature of emotions. American Scientist, 89(4), 344–350.
Real Academia Española. (2010). Ortografía de la lengua española. Espasa-Calpe.
Štular, B., & Belak, M. (2022). Deep data example: Zbiva, early medieval data set for the Eastern Alps. Research Data Journal for the Humanities and Social Sciences, 7(1), 1–13. https://doi.org/10.1163/24523666-bja10024(s’ouvre dans une nouvelle fenêtre)
The configuration of the REWIND corpus was particularly challenging. The selection of sources was based on several criteria to ensure the research aligned coherently with the framework of decolonial gender theory. Thus, all corpus texts were authored by Latin American women, published in Spanish in book format, and written during their travels to predominantly the same European countries in the 20th century, prior to the Second World War. These authors identified with the feminist cause and advocated for cultural diversity. In this way, Inés Echeverría (1937), Clorinda Matto (1909), Zoila Aurora Cáceres (1911), María Enriqueta Camarillo (1930), and Isabel Carrasquilla (2011), become sociocultural interpreters of European cultural heritage, building bridges between their cultural origin and Europe.
The REWIND dataset aligns with the concept of 'deep data' (Štular & Belak, 2022), as the geo-semantic analysis required a close engagement with the text, with each annotation necessitating interpretation. The workflow for encoding texts in XML-TEI format began with the digitisation of paper travel books to generate PDF files. All texts were sourced from public digital repositories, except for Oasis de Arte, which was requested by the National Library of Spain to be digitised. Subsequently, Optical Character Recognition (OCR) techniques were employed to convert the images into plain text files in TXT format, which then served as the basis for encoding.
Before encoding the texts in XML-TEI, a conceptual analysis was conducted to develop the ontological model based on the CIDOC-CRM proposal. The semantic labelling used was grounded in the classes defined in the ontology, ensuring that the information was correlated and interoperable. The UNESCO typology was applied to catalogue different cultural heritage items, distinguishing between intangible cultural heritage (E28 Conceptual Object) and movable (E22 Human-Made Object) and immovable (E53 Place) material cultural heritage. Additionally, the ROSSIO Thesaurus was used to categorise the different elements of European cultural heritage. This classification aligned with controlled vocabulary, enhancing integration and enabling the use of Linked Open Data (LOD). Controlled vocabularies were also employed to limit the meaning of terms used to define an object, addressing natural language issues such as synonymy or polysemy.
The REWIND dataset was created by encoding travel books using XML markup language and the standards proposed by the Text Encoding Initiative. On the one hand, a modernised diplomatic digital edition of each publication was produced, ensuring that not only the structure of the books was annotated, but also that the original formatting of the text (e.g. italics, capital letters, or indentation) and any typographical errors were preserved. Additionally, the Spanish language was also adapted to the most recent edition of the Ortografía de la Lengua Española (RAE, 2010), as the sentiment analysis relied on a contemporary Spanish Emotion Lexicon. On the other hand, tagging was used to identify people, places, objects, and events, turning each book into a database that could be queried using XQuery expressions. This annotation system enabled differentiation of all cultural heritage elements by linking them, whenever possible, with geographical coordinates through the GeoNames gazetteer and to LOD from VIAF and Wikidata. Creating semantically well-structured data from diverse sources was a complex task. To address this, all information was centralised in an index that referenced the same object across different travel books by means of a unique identifier, implemented through the @xml:id attribute.
For elements of material cultural heritage, the
Outcome and results: Comino Comino, A. (2024). REWIND Corpus (1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.14093621(s’ouvre dans une nouvelle fenêtre)
WP2 Geo-semantic annotation of travel books
Historical narrative sources are often characterised by numerous interconnected references to persons, places, institutions, events, dates, etc. For this reason, a knowledge graph based on the CIDOC-CRM (v. 7.1.3 February 2023), a model developed by ICOM and converted into the ISO 21127:2023 standard, was chosen to structure historical information in a manner compatible with the semantic web, rather than using a relational database. The CIDOC-CRM consists of 81 classes and 160 properties, structured hierarchically to represent concepts and information related to cultural heritage and museum documentation. Once the text was encoded in a machine-readable format, it became possible to extract data relating to the elements of European cultural heritage mentioned in the travel books and integrate them into the ontological model. The REWIND project focused on event instances when modelling cultural heritage data contained in travel books, with attention given to what is observed, when, where, and by whom.
An ontology provides a model for the management and semantic representation of knowledge, defining a structured set of concepts and describing the relationships between them through classes and properties, to which values are associated. While humans cannot easily detect interconnections between places, objects, events, and people, digital approaches enable quantification and pattern recognition. Additionally, an ontology allows for linking the historical content intrinsic to the text with external information through the use of LOD. In this context, CIDOC-CRM was used, as the heritage conception in the analysed travel books had a museographic character.
In the REWIND project, a workflow was implemented consisting of an initial phase to identify elements of European cultural heritage mentioned in travel books through XML-TEI encoding, followed by a second phase in which these data were exported to the ontological model for analysis. Once the text was encoded in a machine-readable format using XML-TEI, information related to the elements of European cultural heritage referenced by the authors in their travel books was extracted and integrated into the ontological model. To access the XML documents, the open-source software eXist-db was used. The XQuery language facilitated the browsing of XML-TEI files to locate all heritage references and transfer them to a knowledge graph created with Protégé, a free open-source ontology editor and framework for building intelligent systems developed by Stanford University. This process was relatively straightforward because the semantic labelling employed by XML-TEI was based on the classes and properties defined in the ontology, ensuring that the information was both correlated and interoperable.
Outcome and results: Comino Comino, A. (2024). REWIND Ontology. Zenodo. https://doi.org/10.5281/zenodo.14577943(s’ouvre dans une nouvelle fenêtre)
WP3 Opinion mining and comparative analysis
In a memory space, diverse narratives coexist and compete to shape how the past is represented or recounted, frequently giving rise to contradictions, conflicts, and tensions. Several elements aligned with the concept of dissonant heritage, such as monuments associated with the European colonisation of the Americas, were selected to compare the authors' narratives with authorised heritage discourse. Specifically, in the case of Spain, texts by intellectuals such as Rafael Altamira (1900-1911) and Emilio Castelar (1892), alongside references from travel guides by Baedeker (1898), Bradshaw (1894), and Murray (1890), were analysed to explore discrepancies between official and hegemonic historical narratives and the reception of colonial monuments by Latin American women writers. The aim was to carry out a reverse reading; that is, to examine whether the authorised heritage discourse present in history books and reflected in monuments was also transmitted to travel books through the reception process by travellers. Furthermore, the objective was to assess the influence of the travel guides consulted by the authors on their interpretations, considering that the REWIND project sought alternative testimonies for constructing historical knowledge about cultural heritage.
On the one hand, to contrast discourses from the perspective of otherness, keywords such as 'our', 'hero', 'fatherland', 'distinguished man', and 'conquest' were selected, and their semantic contexts were analysed. On the other hand, sentiment analysis was applied to descriptions of dissonant heritage elements, as emotional responses to heritage during travel have been associated with psychosomatic reactions, including Stendhal, Jerusalem, and Kinkaku-ji syndromes. Starting from the premise that most cultural objects are designed to provoke an aesthetic response, travel books provide valuable resources for sentiment analysis. However, in literary texts, emotions function as an expressive device aimed at eliciting a response from the reader. Therefore, when analysing travel narratives, it was necessary to distinguish between the affective dimension associated with autobiographical narrative and the sensations evoked during reading. Nonetheless, the study of emotions proves highly complex, as their expression is contingent upon the historical context. It was thus essential to consider the psychological theories and sociocultural environments of the period. Using this approach, the REWIND project addressed questions such as: Which heritage elements were selected in travel books? What power relations did they perpetuate? What versions of memory, identity, or sense of place did they legitimise?
Analysing documents in Spanish posed a challenge, as most digital sentiment analysis applications and methods had been developed for English texts. Methodologies, workflows or techniques needed to be adapted. The project adopted a proposal that utilised the R programming language, the Tidytext and Syuzhet packages, and the NRC Emotion Lexicon (Isasi, 2021) to detect and evaluate the emotions present in Spanish-language travel books. The NRC Emotion Lexicon (Mohammad & Turney, 2013) contains 14,182 unigrams manually annotated with both positive or negative sentiment valences and the emotions of anger, anticipation, disgust, fear, joy, sadness, surprise, and trust. Although this procedure was a straightforward approach to the study of the eight 'basic emotions' defined by Plutchik (2001), it allowed for the observation of a connection between heritage and emotionality.
Outcome and results: A set of scripts in R was developed to analyse the texts, but they were ultimately not shared in the project's Zenodo repository due to the identification of biases that were not intended to be perpetuated. Sentiment analysis applied to texts in Spanish proved highly complex, particularly as the REWIND project focused on early 20th century Spanish-language texts from various Spanish-speaking countries. While the expression of emotions is inherently personal, it is also shaped by cultural codes linked to social status and the evolution of language. These factors posed significant challenges when attempting to model emotions associated with descriptive texts about cultural heritage elements. To address these biases, a lexical collocation analysis was conducted to explore associations between adjectives with positive or negative valences and cultural heritage objects or historical figures. Furthermore, the analysis accounted for context by incorporating the effects of negation adverbs.
Bibliography:
Altamira, R. (1900–1911). Historia de España y de la civilización española. Juan Gili.
Baedeker, K. (1898). Spain and Portugal. Karl Baedeker.
Cáceres, A. (1911). Oasis de arte. Garnier Hermanos.
Camarillo, M. E. (1930). Brujas, Lisboa, Madrid. Espasa Calpe.
Carrasquilla, I. (2011). Impresiones de viaje escritas por una abuela para sus nietos. Universidad EAFIT.
Castelar, E. (1892). Historia del descubrimiento de América. [s.n.].
Charnock, R. S. (1894). Bradshaw's illustrated hand-book to Spain and Portugal: A complete guide for travellers in the Peninsula. W. J. Adams & Sons.
Echeverría, I. (1937). Entre dos siglos. Ediciones Ercilla.
Ford, R. (1890). Handbook for travellers in Spain (7th ed.). J. Murray.
Isasi, J. (2021). Análisis de sentimientos en R con ‘syuzhet’. Programming Historian en español, 5. https://doi.org/10.46430/phes0051(s’ouvre dans une nouvelle fenêtre)
Matto, C. (1909). Viaje de recreo: España, Francia, Inglaterra, Italia, Suiza, Alemania. F. Sempere.
Mohammad, S. & Turney, P. D. (2013). Crowdsourcing a Word–Emotion Association Lexicon. Computational Intelligence, 29(3), 436–465. https://doi.org/10.1111/j.1467-8640.2012.00460.x(s’ouvre dans une nouvelle fenêtre)
Plutchik, R. (2001). The nature of emotions. American Scientist, 89(4), 344–350.
Real Academia Española. (2010). Ortografía de la lengua española. Espasa-Calpe.
Štular, B., & Belak, M. (2022). Deep data example: Zbiva, early medieval data set for the Eastern Alps. Research Data Journal for the Humanities and Social Sciences, 7(1), 1–13. https://doi.org/10.1163/24523666-bja10024(s’ouvre dans une nouvelle fenêtre)
Progrès au-delà de l’état des connaissances et impact potentiel prévu (y compris l’impact socio-économique et les conséquences sociétales plus larges du projet jusqu’à présent)
The primary scientific contribution of the REWIND project lies in the multimodal interpretation of travel books as historical sources, supported by a Digital Humanities methodology. This approach represented a paradigm shift in working with written texts, differing significantly from the traditional 'archive' method typically employed in historical sciences and promoting hybrid profiles in the Humanities. The combination of close reading techniques (geo-semantic annotation) and distant reading methods (sentiment analysis, named entity recognition, GIS, semantic analysis, LOD, etc.) revealed the vast potential for extracting new information from digital-format data. From this perspective, it became possible to transform the geographic information contained in travel books into maps that reconstructed the itineraries undertaken by the authors, geolocating each of the heritage elements they mentioned. Furthermore, these maps were either static, such as those created using QGIS, an open-source Geographic Information System software, or dynamic, such as the interactive maps developed with the StoryMapJS application, a free tool developed by the Knight Lab. These diverse methods of exploring the spatial dimension of travel literature, when combined with temporal coordinates, also facilitated the observation of changes in cultural heritage over time, such as physical structural alterations or relocations. These changes reflected the evolving meanings ascribed to heritage, which were perceived differently depending on factors such as gender, ethnicity, status, or religion, and were thus continuously subject to scrutiny and transformation.
The REWIND project also highlighted the importance of constructing polyphonic historical narratives, emphasising that there is no single historical account. From this perspective, it demonstrated how incorporating the perspective of otherness when addressing issues related to dissonant cultural heritage contributes to rethinking public spaces and the ways in which knowledge is created, not only in the 20th century but also in the present day.
The workflow of the REWIND project, documented in various guidelines, is easily replicable as it follows the FAIR principles, uses open-source software, and promotes Open Science. From this perspective, the project does not conclude at this stage, as it was conceived with the understanding that knowledge is the only thing that grows when shared. In addition, both the dataset and the ontological model are also available for other research initiatives in the project's Zenodo repository.
Regarding technological impact, in collaboration with the HD-Lab / ILC – CSIC, a script was created to transform XML data into HTML format and produce interactive digital editions that enhance the use of LOD. Furthermore, work is currently underway to develop an interactive graph based on the ontological model of the REWIND project, built on the ROSSIO Infrastructure.
The REWIND project held significant social impact within the framework of European Union policies as it revisited the authorised heritage discourse reflected in monuments, historic buildings, works of art, and other manifestations of European cultural heritage present in the urban landscapes of many European cities. It encouraged a responsible engagement with these spaces while also promoting alternative tourism initiatives inspired by the itineraries of the studied authors. Furthermore, the REWIND project was intersectoral, enabling the integration of an academic knowledge transfer initiative with products aligned to sectors such as entertainment, education, culture, and tourism. The project website offers a navigation experience through audio recordings, texts, interactive maps, and other resources designed to provide a geo-textual immersive adventure, either on-site or from home. This approach allows individuals who are unable to travel for various reasons to explore European cultural heritage. Additionally, it contributes to reducing the carbon footprint, as one can journey alongside Echeverría, Matto, Cáceres, Camarillo, and Carrasquilla through the interactive digital edition developed within the project, embracing a perspective of otherness. Consequently, this action continues to contribute to the improvement of society by recovering the memory of past non-European women, following the United Nations' Sustainable Development Goals for 2030: SDG 5 Gender equality, SDG 11 Sustainable cities and communities, and SDG 16 Peace, justice, and strong institutions. Thus, the REWIND project constituted an analytical tool to develop inclusive cultural policies that favour the image of European cities, the identification of citizens with heritage, the defence of an egalitarian society, and the promotion of sustainable tourism.
The REWIND project also highlighted the importance of constructing polyphonic historical narratives, emphasising that there is no single historical account. From this perspective, it demonstrated how incorporating the perspective of otherness when addressing issues related to dissonant cultural heritage contributes to rethinking public spaces and the ways in which knowledge is created, not only in the 20th century but also in the present day.
The workflow of the REWIND project, documented in various guidelines, is easily replicable as it follows the FAIR principles, uses open-source software, and promotes Open Science. From this perspective, the project does not conclude at this stage, as it was conceived with the understanding that knowledge is the only thing that grows when shared. In addition, both the dataset and the ontological model are also available for other research initiatives in the project's Zenodo repository.
Regarding technological impact, in collaboration with the HD-Lab / ILC – CSIC, a script was created to transform XML data into HTML format and produce interactive digital editions that enhance the use of LOD. Furthermore, work is currently underway to develop an interactive graph based on the ontological model of the REWIND project, built on the ROSSIO Infrastructure.
The REWIND project held significant social impact within the framework of European Union policies as it revisited the authorised heritage discourse reflected in monuments, historic buildings, works of art, and other manifestations of European cultural heritage present in the urban landscapes of many European cities. It encouraged a responsible engagement with these spaces while also promoting alternative tourism initiatives inspired by the itineraries of the studied authors. Furthermore, the REWIND project was intersectoral, enabling the integration of an academic knowledge transfer initiative with products aligned to sectors such as entertainment, education, culture, and tourism. The project website offers a navigation experience through audio recordings, texts, interactive maps, and other resources designed to provide a geo-textual immersive adventure, either on-site or from home. This approach allows individuals who are unable to travel for various reasons to explore European cultural heritage. Additionally, it contributes to reducing the carbon footprint, as one can journey alongside Echeverría, Matto, Cáceres, Camarillo, and Carrasquilla through the interactive digital edition developed within the project, embracing a perspective of otherness. Consequently, this action continues to contribute to the improvement of society by recovering the memory of past non-European women, following the United Nations' Sustainable Development Goals for 2030: SDG 5 Gender equality, SDG 11 Sustainable cities and communities, and SDG 16 Peace, justice, and strong institutions. Thus, the REWIND project constituted an analytical tool to develop inclusive cultural policies that favour the image of European cities, the identification of citizens with heritage, the defence of an egalitarian society, and the promotion of sustainable tourism.