SuCCESS’16 - First International Workshop on Semantic Change & Evolving Semantics
The evolution and change of semantics is central to the objectives of the PERICLES project: semantics relates to our shared understanding of the world around us, and in cases where there is a mismatch, then this can lead to a breakdown in our shared understanding and an inability to communicate and exchange information effectively. A couple of well-known examples might be US versus UK English, where in the former you would use the terms “elevator” and “sidewalk” to refer to “lift” and “pavement” in the latter. Generally speaking, these differences are well known and can be understood from their context. However in the case of digital preservation, which is the focus of the PERICLES project, such contextual information to aid interpretation is often absent (or has itself been subject to change), and therefore presents more of a challenge for the retrieval and interpretation of information and digital objects. However, it is not an interest that is limited to the project – or indeed to preservation – but a topic relevant to many disciplines and areas of research.
This was reflected in the variety of presenters and participants involved in the recent SuCCESS’16 - First International Workshop on Semantic Change & Evolving Semantics, which was part of the 2016 SEMANTiCS conference in Leipzig (12-15 September 2016). The seminar room was packed as more than 20 participants (and we suspect one or two more who managed to sneak in) listened to the keynote and invited papers and discussed all things semantic change related.
With the introductions out of the way, the keynote speaker Albert Meroño Peñuela from the Vrije Universiteit Amsterdam kicked the session off with a thought provoking talk which marked out some of the issues relating to semantic change. In particular he was able to draw upon his own experience during the CLARIAH, DataLegend and CEDAR projects, in which he has focused on building a distributed infrastructure for the humanities and social sciences. In particular, one of the key observations he made related to the impact of change in integrating historical statistical data from different domains. The talk ended with a vision of future directions for the study of semantic change, especially regarding the combination of statistical and symbolic representations, the need of new vocabularies, and the opportunity of version control systems and new datasets.
The rest of the workshop was divided into two groups of paper presentations: The first group session focused on studies of semantic change, and the second group session concentrated more on wider investigations and applications of this research.
The first session presented a number of interesting approaches for studying semantic change. For a start, a paper from members of the University of Leipzig adopted context volatility surrounding terms as a way of measuring their change in usage (and meaning). One particularly interesting innovation was the adoption of approaches from recommender systems to overcome the sparseness of language in order to study change in words and topics in newspapers. The final two papers in this group were presented by PERICLES project partners from the University of Boras and CERTH: The first focused on using approaches from physics to describe semantic drift, and in particular the Somoclu tool developed at the University of Borås. Since Somoclu has been publicly released (see this post for greater description), it was good to learn that one of the comments following the presentation was from one of the other speakers who has already used Somoclu on their own data (and – happily! – found it easy to use). The final paper of the session explored the specific case of measuring semantic drift in ontologies via ontology evolution, and provided an interesting insight into how concepts/labels of art works in the Tate Galleries catalogues changed over a ten year period as technologies adopted by artists became more specialised.
The second session of papers took a broader, more applied view of change. The first paper by a team from Saarland University and the Research Institute for Linguistics of the Hungarian Academy of Sciences used Twitter data to track the factuality of rumours using lexical cues in relation to crisis events. The second paper by PERICLES project partner KCL along with the University of Bristol also uses data from social media, in this case to study community change over time using dynamic social network analysis and topic modelling. The final paper of the session from the French Space Agency (Centre National d'Études Spatiales) described a large body of work which had considered semantic shift in order to detect knowledge changes. It discussed the contribution of semiotics to that issue. As this description of the papers shows, there was a large variety of perspectives brought to this workshop on semantic change, both in terms of the research backgrounds of the presenters, as well as the specific area of change that they have investigated. We also acknowledge the involvement of the audience who really contributed to the discussions with questions and comments. For those interested in learning more about this topic and the workshop, the papers from the workshop can be found in the CEUR Workshop Proceedings website (see session 2) and we will also draw together and review the themes and challenges of this topic in a separate report.
Although we have noted some advances in research studying semantic evolution and change, the workshop has not ‘solved’ the issue entirely in preservation - or more broadly - in any other discipline. On the one hand, this is quite a good thing because it means that we can still study and research this fascinating topic (and also organise future SuCCESS workshops!). However, what the workshop has done is build upon the work of the PERICLES Community of Practice on Evolving Semantics, which has been meeting to discuss work in this area over the past 12 months; in addition it has also broadened participation in these discussion to a wider range of researchers coming from an even wider range of disciplines. It is through the combined knowledge, perspectives and insights of this group that we can develop a better understanding of this fascinating topic, and ultimately mitigate the challenges of such change in digital preservation – this is something we have done in the PERICLES project.
On behalf of the other organisers and presenters, we wish to thank everyone involved in the First International Workshop on Semantic Change & Evolving Semantics, and look forward to discussing more on the topic in the future.