kth.sePublications
Change search
Link to record
Permanent link

Direct link
Publications (10 of 64) Show all publications
Cornell, F., Karlgren, J., Sachan, A. & Girdzijauskas, S. (2022). Symbolic Hyperdimensional Vectors with Sparse Graph Convolutional Neural Networks. In: 2022 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN): . Paper presented at IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) / IEEE World Congress on Computational Intelligence (IEEE WCCI) / International Joint Conference on Neural Networks (IJCNN) / IEEE Congress on Evolutionary Computation (IEEE CEC), JUL 18-23, 2022, Padua, ITALY. Institute of Electrical and Electronics Engineers (IEEE)
Open this publication in new window or tab >>Symbolic Hyperdimensional Vectors with Sparse Graph Convolutional Neural Networks
2022 (English)In: 2022 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), Institute of Electrical and Electronics Engineers (IEEE) , 2022Conference paper, Published paper (Refereed)
Abstract [en]

In this paper, we propose a novel way of representing graphs for processing in Graph Neural Networks. We reduce the dimensionality of the input data by using Random Indexing, a Vector Symbolic Architectural framework; we implement a new trainable neural layer, also inspired by Vector Symbolic Architectures; we leverage the sparseness of the incoming data in a Sparse Neural Network framework. Our experiments on a number of publicly available datasets and standard benchmarks demonstrate that we can reduce the number of parameters by up to two orders of magnitude. We show how this parsimonious approach not only delivers competitive results but even improves performance for node classification and link prediction. We find that this holds in particular for cases where the graph lacks node features.

Place, publisher, year, edition, pages
Institute of Electrical and Electronics Engineers (IEEE), 2022
Series
IEEE International Joint Conference on Neural Networks (IJCNN), ISSN 2161-4393
Keywords
vector symbolic architectures, graph neural networks, random indexing
National Category
Computer Sciences
Identifiers
urn:nbn:se:kth:diva-323028 (URN)10.1109/IJCNN55064.2022.9892300 (DOI)000867070903060 ()2-s2.0-85140763914 (Scopus ID)
Conference
IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) / IEEE World Congress on Computational Intelligence (IEEE WCCI) / International Joint Conference on Neural Networks (IJCNN) / IEEE Congress on Evolutionary Computation (IEEE CEC), JUL 18-23, 2022, Padua, ITALY
Note

Part of proceedings: ISBN 978-1-7281-8671-9

QC 20230112

Available from: 2023-01-12 Created: 2023-01-12 Last updated: 2023-12-11Bibliographically approved
Karlgren, J. (2019). Adopting systematic evaluation benchmarks in operational settings. In: Nicola Ferro and Carol Peters (Ed.), Information Retrieval Evaluation in a Changing World: Lessons Learned from 20 Years of CLEF (pp. 583-590). Cham: Springer Berlin/Heidelberg
Open this publication in new window or tab >>Adopting systematic evaluation benchmarks in operational settings
2019 (English)In: Information Retrieval Evaluation in a Changing World: Lessons Learned from 20 Years of CLEF / [ed] Nicola Ferro and Carol Peters, Cham: Springer Berlin/Heidelberg, 2019, p. 583-590Chapter in book (Refereed)
Abstract [en]

Evaluationofinformationsystemsincommercialandindustrialsettings differs from academic evaluation of methodology in important ways. Those dif- ferences have to do with differing organisational priorities between practice and research. Some of those priorities can be adjusted, others must be taken into account, to be able to include evaluation into an operational development pipeline.

Place, publisher, year, edition, pages
Cham: Springer Berlin/Heidelberg, 2019
Series
Information Retrieval Series ; 41
Keywords
Evaluation, Information retrieval, Use case
National Category
Information Systems
Identifiers
urn:nbn:se:kth:diva-286346 (URN)10.1007/978-3-030-22948-1_25 (DOI)
Note

QC 20201126

Available from: 2020-11-25 Created: 2020-11-25 Last updated: 2024-03-18Bibliographically approved
Karlgren, J. & Kanerva, P. (2019). High-dimensional distributed semantic spaces for utterances. Natural Language Engineering, 25(4), 503-517
Open this publication in new window or tab >>High-dimensional distributed semantic spaces for utterances
2019 (English)In: Natural Language Engineering, ISSN 1351-3249, E-ISSN 1469-8110, Vol. 25, no 4, p. 503-517Article in journal (Refereed) Published
Abstract [en]

High-dimensional distributed semantic spaces have proven useful and effective for aggregating and processing visual, auditory and lexical information for many tasks related to human-generated data. Human language makes use of a large and varying number of features, lexical and constructional items as well as contextual and discourse-specific data of various types, which all interact to represent various aspects of communicative information. Some of these features are mostly local and useful for the organisation of, for example, argument structure of a predication; others are persistent over the course of a discourse and necessary for achieving a reasonable level of understanding of the content. This paper describes a model for high-dimensional representation for utterance and text-level data including features such as constructions or contextual data, based on a mathematically principled and behaviourally plausible approach to representing linguistic information. The implementation of the representation is a straightforward extension of Random Indexing models previously used for lexical linguistic items. The paper shows how the implementedmodel is able to represent a broad range of linguistic features in a common integral framework of fixed dimensionality, which is computationally habitable, and which is suitable as a bridge between symbolic representations such as dependency analysis and continuous representations used, for example, in classifiers or further machine-learning approaches. This is achieved with operations on vectors that constitute a powerful computational algebra, accompanied with an associative memory for the vectors. The paper provides a technical overview of the framework and a worked through implemented example of how it can be applied to various types of linguistic features.

Place, publisher, year, edition, pages
Cambridge University Press, 2019
Keywords
constructional grammar, High-dimensional computing, random indexing
National Category
Other Computer and Information Science
Identifiers
urn:nbn:se:kth:diva-262607 (URN)10.1017/S1351324919000226 (DOI)000477972600006 ()2-s2.0-85070086799 (Scopus ID)
Note

QC 20191017

Available from: 2019-10-17 Created: 2019-10-17 Last updated: 2022-06-26Bibliographically approved
Karlgren, J. (2019). How Lexical Gold Standards Have Effects on the Usefulness of Text Analysis Tools for Digital Scholarship. In: Experimental IR Meets Multilinguality, Multimodality, and Interaction: 10th International Conference of the CLEF Association, CLEF 2019, Lugano, Switzerland, September 9–12, 2019, Proceedings. Paper presented at 10th International Conference of the CLEF Association, CLEF 2019, Lugano, Switzerland, September 9–12, 2019 (pp. 178-184). Springer, 11696
Open this publication in new window or tab >>How Lexical Gold Standards Have Effects on the Usefulness of Text Analysis Tools for Digital Scholarship
2019 (English)In: Experimental IR Meets Multilinguality, Multimodality, and Interaction: 10th International Conference of the CLEF Association, CLEF 2019, Lugano, Switzerland, September 9–12, 2019, Proceedings, Springer, 2019, Vol. 11696, p. 178-184Conference paper, Published paper (Refereed)
Abstract [en]

This paper describes how the current lexical similarity and analogy gold standards are built to conform to certain ideas about what the models they are designed to evaluate are used for. Topical relevance has always been the most important target notion for information access tools and related language technology technologies, and while this has proven a useful starting point for much of what information technology is used for, it does not always align well with other uses to which technologies are being put, most notably use cases from digital scholarship in the humanities or social sciences. This paper argues for more systematic formulation of requirements from the digital humanities and social sciences and more explicit description of the assumptions underlying model design.

Place, publisher, year, edition, pages
Springer, 2019
Series
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), ISSN 0302-9743
National Category
Language Technology (Computational Linguistics)
Identifiers
urn:nbn:se:kth:diva-268334 (URN)10.1007/978-3-030-28577-7_14 (DOI)000611683100014 ()2-s2.0-85072842110 (Scopus ID)9783030285760 (ISBN)
Conference
10th International Conference of the CLEF Association, CLEF 2019, Lugano, Switzerland, September 9–12, 2019
Note

QC 20200701

Available from: 2020-03-10 Created: 2020-03-10 Last updated: 2022-06-26Bibliographically approved
Fano, E., Karlgren, J. & Nivre, J. (2019). Uppsala University and Gavagai at CLEF Erisk: Comparing word embedding models. In: CEUR Workshop Proceedings: . Paper presented at 20th Working Notes of CLEF Conference and Labs of the Evaluation Forum, CLEF 2019, 9 September 2019 through 12 September 2019. CEUR-WS, 2380
Open this publication in new window or tab >>Uppsala University and Gavagai at CLEF Erisk: Comparing word embedding models
2019 (English)In: CEUR Workshop Proceedings, CEUR-WS , 2019, Vol. 2380Conference paper, Published paper (Refereed)
Abstract [en]

This paper describes an experiment to evaluate the performance of three different types of semantic vectors or word embeddings-random indexing, GloVe, and ELMo-and two different classification architectures-linear regression and multi-layer perceptrons-for the specific task of identifying authors with eating disorders from writings they publish on a discussion forum. The task requires the classifier to process texts written by the authors in the sequence they were published, and to identify authors likely to be at risk of suffering from eating disorders as early as possible. The data are part of the eRISK evaluation task of CLEF 2019 and evaluated according to the eRISK metrics. Contrary to our expectations, we did not observe a clear-cut advantage using the recently popular contextualized ELMo vectors over the commonly used and much more light-weight GloVe vectors, or the more handily learnable random indexing vectors.

Place, publisher, year, edition, pages
CEUR-WS, 2019
Keywords
Author classification, Semantic vectors, Word embeddings
National Category
Language Technology (Computational Linguistics)
Identifiers
urn:nbn:se:kth:diva-257945 (URN)2-s2.0-85070497793 (Scopus ID)
Conference
20th Working Notes of CLEF Conference and Labs of the Evaluation Forum, CLEF 2019, 9 September 2019 through 12 September 2019
Note

QC 20190909

Available from: 2019-09-09 Created: 2019-09-09 Last updated: 2022-06-26Bibliographically approved
Espinoza, F., Hamfors, O., Karlgren, J., Olsson, F., Persson, P., Hamberg, L. & Sahlgren, M. (2018). Analysis of Open Answers to Survey Questions throughInteractive Clustering and Theme Extraction. In: Proceedings of Conference on Human Information Interaction & Retrieval: . Paper presented at CHIIR (pp. 317-320). ACM Digital Library
Open this publication in new window or tab >>Analysis of Open Answers to Survey Questions throughInteractive Clustering and Theme Extraction
Show others...
2018 (English)In: Proceedings of Conference on Human Information Interaction & Retrieval, ACM Digital Library, 2018, p. 317-320Conference paper, Published paper (Refereed)
Abstract [en]

This paper describes design principles for and the implementation of Gavagai Explorer—a new application which builds on interactive text clustering to extract themes from topically coherent text sets such as open text answers to surveys or questionnaires.An automated system is quick, consistent, and has full coverage over the study material. A system allows an analyst to analyze more answers in a given time period; provides the same initial results regardless of who does the analysis, reducing the risks of inter-rater discrepancy; and does not risk miss responses due to fatigue or boredom. These factors reduce the cost and increase the reliability of the service. The most important feature, however, is relieving the human analyst from the frustrating aspects of the coding task, freeing the effort to the central challenge of understanding themes. Gavagai Explorer is available on-line at http://explorer.gavagai.se

Place, publisher, year, edition, pages
ACM Digital Library, 2018
Keywords
Online analytical processing, User interface design
National Category
Human Computer Interaction
Research subject
Computer Science
Identifiers
urn:nbn:se:kth:diva-223645 (URN)10.1145/3176349.3176892 (DOI)000460388100049 ()2-s2.0-85050455594 (Scopus ID)
Conference
CHIIR
Note

QC 20180306

Available from: 2018-02-27 Created: 2018-02-27 Last updated: 2022-06-26Bibliographically approved
Karlgren, J., Esposito, L., Gratton, C. & Kanerva, P. (2018). Authorship profiling without using topical information: Notebook for PAN at CLEF 2018. In: CLEF 2018 Working Notes: . Paper presented at 19th Working Notes of CLEF Conference and Labs of the Evaluation Forum, CLEF 2018, Avignon, France, 10 September 2018 through 14 September 2018. CEUR-WS, 2125
Open this publication in new window or tab >>Authorship profiling without using topical information: Notebook for PAN at CLEF 2018
2018 (English)In: CLEF 2018 Working Notes, CEUR-WS , 2018, Vol. 2125Conference paper, Published paper (Refereed)
Abstract [en]

This paper describes an experiment made for the PAN 2018 shared task on author profiling. The task is to distinguish female from male authors of microblog posts published on Twitter using no extraneous information except what is in the posts; this experiment focusses on using non-topical information from the posts, rather than gender differences in referential content.

Place, publisher, year, edition, pages
CEUR-WS, 2018
Series
CEUR Workshop Proceedings, ISSN 1613-0073 ; 2125
National Category
Gender Studies
Identifiers
urn:nbn:se:kth:diva-234054 (URN)2-s2.0-85051071094 (Scopus ID)
Conference
19th Working Notes of CLEF Conference and Labs of the Evaluation Forum, CLEF 2018, Avignon, France, 10 September 2018 through 14 September 2018
Funder
VINNOVA
Note

QC 20180905

Available from: 2018-09-05 Created: 2018-09-05 Last updated: 2022-06-26Bibliographically approved
Karlgren, J. & Kanerva, P. (2018). Hyperdimensional utterance spaces. In: CEUR Workshop Proceedings: . Paper presented at 1st Biennial Conference on Design of Experimental Search and Information Retrieval Systems, DESIRES 2018, Bertinoro, Italy, 28 August 2018 through 31 August 2018 (pp. 29-35). CEUR-WS, 2167
Open this publication in new window or tab >>Hyperdimensional utterance spaces
2018 (English)In: CEUR Workshop Proceedings, CEUR-WS , 2018, Vol. 2167, p. 29-35Conference paper, Published paper (Refereed)
Abstract [en]

Human language has a large and varying number of features, both lexical items and constructions, which interact to represent various aspects of communicative information. High-dimensional semantic spaces have proven useful and effective for aggregating and processing lexical information for many language processing tasks. This paper describes a hyperdimensional processing model for language data, a straightforward extension of models previously used for words to handling utterance or text level information. A hyperdimensional model is able to represent a broad range of linguistic and extra-linguistic features in a common integral framework which is suitable as a bridge between symbolic and continuous representations, as an encoding scheme for symbolic information and as a basis for feature space exploration. This paper provides an overview of the framework and an example of how it is used in a pilot experiment.

Place, publisher, year, edition, pages
CEUR-WS, 2018
Series
CEUR Workshop Proceedings, ISSN 1613-0073
Keywords
Constructional grammar, Hyperdimensional computing, Knowledge representation and reasoning, Utterance-level semantics
National Category
Language Technology (Computational Linguistics)
Identifiers
urn:nbn:se:kth:diva-234755 (URN)2-s2.0-85052649405 (Scopus ID)
Conference
1st Biennial Conference on Design of Experimental Search and Information Retrieval Systems, DESIRES 2018, Bertinoro, Italy, 28 August 2018 through 31 August 2018
Note

QC 20180911

Available from: 2018-09-11 Created: 2018-09-11 Last updated: 2022-06-26Bibliographically approved
Karlgren, J. (2018). Regulation of Unpredictable Effects of Decision Making Systems is Non-trivial. In: Peter Wahlgren (Ed.), 50 Years of Law and IT: The Swedish Law and Informatics Research Institute 1968-2018: (pp. 127-132). Stockholm: The Stockholm University Law Faculty
Open this publication in new window or tab >>Regulation of Unpredictable Effects of Decision Making Systems is Non-trivial
2018 (English)In: 50 Years of Law and IT: The Swedish Law and Informatics Research Institute 1968-2018 / [ed] Peter Wahlgren, Stockholm: The Stockholm University Law Faculty , 2018, p. 127-132Chapter in book (Other academic)
Abstract [en]

Technical advances are rapidly delegating decision making in newarenas of human activity to information systems through theapplication of new classification mechanisms from machine learningresearch. How to manage technology-induced change and its effectsthrough legislative systems in order to encourage and supportbehaviour and activities which is desirable and beneficial to thepublic good and dissuade from such which is not is non-trivial. Ingeneral, legislation to cover new technical advances will be based onexisting technology and existing practice. This may seen reasonablebasis to build from and adds legitimacy to regulation and itsapplication, but regulation of technology too often stumbles at thebalancing line between under- standing and promoting future changeproductively and protecting past practice. This paper argues that morethought must be put into the aims of regulatory activities.

Place, publisher, year, edition, pages
Stockholm: The Stockholm University Law Faculty, 2018
Series
Scandinavian Studies in Law, ISSN 0085-5944 ; 65
Keywords
decision making support, machine learning, articificial intelligence, regulatory frameworks, legal implications
National Category
Law and Society
Research subject
Computer Science
Identifiers
urn:nbn:se:kth:diva-238616 (URN)
Note

QC 20181106

Available from: 2018-11-05 Created: 2018-11-05 Last updated: 2022-06-26Bibliographically approved
Iatropoulos, G., Herman, P., Lansner, A., Karlgren, J., Larsson, M. & Olofsson, J. K. (2018). The language of smell: Connecting linguistic and psychophysical properties of odor descriptors. Cognition, 178, 37-49
Open this publication in new window or tab >>The language of smell: Connecting linguistic and psychophysical properties of odor descriptors
Show others...
2018 (English)In: Cognition, ISSN 0010-0277, E-ISSN 1873-7838, Vol. 178, p. 37-49Article in journal (Refereed) Published
Abstract [en]

The olfactory sense is a particularly challenging domain for cognitive science investigations of perception, memory, and language. Although many studies show that odors often are difficult to describe verbally, little is known about the associations between olfactory percepts and the words that describe them. Quantitative models of how odor experiences are described in natural language are therefore needed to understand how odors are perceived and communicated. In this study, we develop a computational method to characterize the olfaction-related semantic content of words in a large text corpus of internet sites in English. We introduce two new metrics: olfactory association index (OAI, how strongly a word is associated with olfaction) and olfactory specificity index (OSI, how specific a word is in its description of odors). We validate the OAI and OSI metrics using psychophysical datasets by showing that terms with high OAI have high ratings of perceived olfactory association and are used to describe highly familiar odors. In contrast, terms with high OSI have high inter-individual consistency in how they are applied to odors. Finally, we analyze Dravnieks's (1985) dataset of odor ratings in terms of OAI and OSI. This analysis reveals that terms that are used broadly (applied often but with moderate ratings) tend to be olfaction-unrelated and abstract (e.g., “heavy” or “light”; low OAI and low OSI) while descriptors that are used selectively (applied seldom but with high ratings) tend to be olfaction-related (e.g., “vanilla” or “licorice”; high OAI). Thus, OAI and OSI provide behaviorally meaningful information about olfactory language. These statistical tools are useful for future studies of olfactory perception and cognition, and might help integrate research on odor perception, neuroimaging, and corpus-based linguistic models of semantic organization.

Place, publisher, year, edition, pages
Elsevier, 2018
Keywords
Computational linguistics, Distributional semantics, Odour identification, Odour naming, Sensory lexicon, Sensory-semantic integration
National Category
Language Technology (Computational Linguistics)
Identifiers
urn:nbn:se:kth:diva-228700 (URN)10.1016/j.cognition.2018.05.007 (DOI)000439402400004 ()29763790 (PubMedID)2-s2.0-85047188460 (Scopus ID)
Note

QC 20180530

Available from: 2018-05-30 Created: 2018-05-30 Last updated: 2022-06-26Bibliographically approved
Organisations
Identifiers
ORCID iD: ORCID iD iconorcid.org/0000-0003-4042-4919

Search in DiVA

Show all publications