Concept Mining with Self-Organizing Maps for the Semantic Web

Honkela, Timo; Pöllä, Matti

doi:10.1007/978-3-642-02397-2_12

Cited by 7 publications

(7 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There is also previous work on concept mining for the Semantic Web with SOM, for example the work by Honkela et al [11], where the emergent structure of an organized SOM reflects the structure of underlying information, used in the training process. Their research also shows that multiple layers of superclass layers can be seen as different-sized nested zones on the SOM grid.…”

Section: Related Work and Development Discussionmentioning

confidence: 99%

Building a Finnish SOM-based ontology concept tagger and harvester

Nyrkkö

2018

Proceedings of the Fourth International Workshop on Computatinal Linguistics of Uralic Languages

View full text Add to dashboard Cite

I demonstrate here an experiment of word sense disambiguation method based on the Self-Organizing Map (SOM) and a pre-existing set of tools for analyzing text in Finnish. It is given a Semantic Web ontology as a reference model, and a related Finnish text corpus with sample term tagging related to the ontology concepts. The experiment is based on "OntoR", a previous experiment on SOM-based ontology term tagging for English. In this work the OntoR model is adapted to the Finnish language, and it is trained on a small text example with hand-picked concept annotations. This computational model can be considered useful for Information Retrieval and concept harvesting purposes in a specific domain where a limited training data set is available. The model adapted to Finnish text analysis stands on OMORFI and HFST morphological analysis, and uses the SOM-PAK library for unsupervised clustering, and ontology concept tagging and further for concept harvesting in Semantic Web ontology development. TiivistelmäKehitän luonnollisessa kielessä ilmenevien sanojen merkitysten erotteluun sopivaa automaattista koneoppivaa työkalua. Laskennallinen malli perustuu itseoppivaan karttaan (SOM, Self-Organizing Map) ja annettuun suomenkieliseen semanttisen webin ontologiaan. Malli oppii tunnistamaan käsitteiden ilmenemistä mallitekstistä, johon on annotoitu (tagattu) malliksi aiemmin laaditun ongologian käsitteitä. Koe liittyy aiemmin englanninkielisten käsitteiden taggaamiseen liittyvään OntoR-koejärjestelyyn joka tutki tekstisyötteessä ilmenevien termien liittämistä SOM-kartan soluihin malliksi annetun annotoidun tekstiesimerkin avulla. Tällainen malli oppii annetun käsitemallin huomattavan niukalla esimerkkiaineistolla ja sopii käyttökohteisiin joissa ei ole tarjolla riittävän suurta datamäärää syvän oppimisen neuroverkkomallin opettamiseksi. Suomenkielisen kokeen morfologisen analyysin pohjalla on OMORFI-ja HFST-työkalut. Koneoppimisen toteuttava SOM-kartta lasketaan SOM-PAK-ohjelmistopaketin avulla. Kehitettyä laskennallista mallia käytetään käsitteiden tunnistamisen lisäksi myös uusien ontologiakäsitteiden ehdokkaiden löytämiseksi.

show abstract

Section: Related Work and Development Discussionmentioning

confidence: 99%

Building a Finnish SOM-based ontology concept tagger and harvester

Nyrkkö

2018

Proceedings of the Fourth International Workshop on Computatinal Linguistics of Uralic Languages

View full text Add to dashboard Cite

show abstract

“…However, in practice the representations created by different people and organizations even in a narrowly defined domain tend to vary. Various semantic mapping techniques have been developed but, in general, it seems necessary to ground semantic representations in their relevant contexts [21], [22]. The GICA method provides a framework in which both grounding in context and modeling subjective or organizational semantic variation can take place.…”

Section: Discussionmentioning

confidence: 99%

Subjects on objects in contexts: Using GICA method to quantify epistemological subjectivity

Honkela

Raitio

Lagus

et al. 2012

The 2012 International Joint Conference on Neural Networks (IJCNN)

Self Cite

View full text Add to dashboard Cite

Abstract-A substantial amount of subjectivity is involved in how people use language and conceptualize the world. Computational methods and formal representations of knowledge usually neglect this kind of individual variation. We have developed a novel method, Grounded Intersubjective Concept Analysis (GICA), for the analysis and visualization of individual differences in language use and conceptualization. The GICA method first employs a conceptual survey or a text mining step to elicit from varied groups of individuals the particular ways in which terms and associated concepts are used among the individuals. The subsequent analysis and visualization reveals potential underlying groupings of subjects, objects and contexts. One way of viewing the GICA method is to compare it with the traditional word space models. In the word space models, such as latent semantic analysis (LSA), statistical analysis of word-context matrices reveals latent information. A common approach is to analyze term-document matrices in the analysis. The GICA method extends the basic idea of the traditional term-document matrix analysis to include a third dimension of different individuals. This leads to a formation of a third-order tensor of size subjects × objects × contexts. Through flattening into a matrix, these subject-object-context (SOC) tensors can again be analyzed using various computational methods including principal component analysis (PCA), singular value decomposition (SVD), independent component analysis (ICA) or any existing or future method suitable for analyzing highdimensional data sets. In order to demonstrate the use of the GICA method, we present the results of two case studies. In the first case, GICA of health-related concepts is conducted. In the second one, the State of the Union addresses by US presidents are analyzed. In these case studies, we apply multidimensional scaling (MDS), the self-organizing map (SOM) and Neighborhood Retrieval Visualizer (NeRV) as specific data analysis methods within the overall GICA method. The GICA method can be used, for instance, to support education of heterogeneous audiences, public planning processes and participatory design, conflict resolution, environmental problem solving, interprofessional and interdisciplinary communication, product development processes, mergers of organizations, and building enhanced knowledge representations in semantic web.

show abstract

“…In one notable application, the WebSOM project indexed over a million web pages and organized them by topical similarity. In another example, the self‐organizing map was used to classify a corpus of about 10,000 scientific abstracts. The largest published project mapped about 7 million patent abstracts to a 1‐million‐node SOM using 500‐dimensional feature vectors.…”

Section: Related Workmentioning

confidence: 99%

“…At another level of analysis, the SOM has been used to cluster words by semantic and syntactic role . Construction of such word category maps has been shown to work well for a wide variety of languages, including flectional (e.g., English), agglutinative (e.g., Finnish), and even tonal isolative (e.g., Chinese) languages .…”

Section: Related Workmentioning

confidence: 99%

“…We have compared our approach to latent semantic analysis using matrix‐algebraic methods, hidden Markov expectation maximization, non‐SOM approaches using TF‐IDF and document weighting factors, and other non‐SOM approaches to word‐co‐occurrence statistics . Finally, we have discussed work similar to our own, like WebSOM and related methods , …”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Self‐organizing maps for latent semantic analysis of free‐form text in support of public policy analysis

Till

Longo

Dobell

et al. 2013

WIREs Data Min & Knowl

View full text Add to dashboard Cite

The huge amount of free-form unstructured text in the blogosphere, its increasing rate of production, and its shrinking window of relevance, present serious challenges to the public policy analyst who seeks to take public opinion into account. Most of the tools which address this problem use XML tagging and other Web 3.0 approaches, which do not address the actual content of blog posts and the associated commentary. We give a tutorial review of latent semantic analysis and the self-organizing maps, as considered in this context, and show how to apply the self-organizing map over a probabilistic latent semantic space to the problem of completely unsupervised clustering of unstructured text in such a way as to be entirely independent of spelling, grammar, and even source language. This provides an algorithm suitable for clustering free-form commentary with a well-structured test environment. The algorithm is applied to academic paper abstracts instead, treated as unstructured text as though they were blog posts, because this set of documents has a known ground truth. The algorithm constructs a word category map and a document map in which words with similar meaning and documents with similar content are clustered together.

show abstract

Concept Mining with Self-Organizing Maps for the Semantic Web

Cited by 7 publications

References 19 publications

Building a Finnish SOM-based ontology concept tagger and harvester

Building a Finnish SOM-based ontology concept tagger and harvester

Subjects on objects in contexts: Using GICA method to quantify epistemological subjectivity

Self‐organizing maps for latent semantic analysis of free‐form text in support of public policy analysis

Contact Info

Product

Resources

About