Limits...
Using Graph Components Derived from an Associative Concept Dictionary to Predict fMRI Neural Activation Patterns that Represent the Meaning of Nouns.

Akama H, Miyake M, Jung J, Murphy B - PLoS ONE (2015)

Bottom Line: We measure the effectiveness of graph-based coefficients through the application of linguistic graph information for a neural activity recorded during conceptual processing in the human brain.Furthermore, correlating the voxel information with the MiF-based principal components, a new computational neurolinguistics model with a network connectivity paradigm is created.This allows two dimensions of context space to be incorporated with both semantic and neural distributional representations.

View Article: PubMed Central - PubMed

Affiliation: Graduate School of Decision Science and Technology, Tokyo Institute of Technology, Tokyo, Japan.

ABSTRACT
In this study, we introduce an original distance definition for graphs, called the Markov-inverse-F measure (MiF). This measure enables the integration of classical graph theory indices with new knowledge pertaining to structural feature extraction from semantic networks. MiF improves the conventional Jaccard and/or Simpson indices, and reconciles both the geodesic information (random walk) and co-occurrence adjustment (degree balance and distribution). We measure the effectiveness of graph-based coefficients through the application of linguistic graph information for a neural activity recorded during conceptual processing in the human brain. Specifically, the MiF distance is computed between each of the nouns used in a previous neural experiment and each of the in-between words in a subgraph derived from the Edinburgh Word Association Thesaurus of English. From the MiF-based information matrix, a machine learning model can accurately obtain a scalar parameter that specifies the degree to which each voxel in (the MRI image of) the brain is activated by each word or each principal component of the intermediate semantic features. Furthermore, correlating the voxel information with the MiF-based principal components, a new computational neurolinguistics model with a network connectivity paradigm is created. This allows two dimensions of context space to be incorporated with both semantic and neural distributional representations.

No MeSH data available.


MDS results for the two distance matrices on the 60 fMRI nouns.We compared the PC-score matrix of MiF-EAT (top) with the co-occurrence matrix of Mitchell et al.’s model from the Google Web 1T 5-gram collection (bottom). The natural numbers attached to the nouns represent the semantic categories (animals: cyan-(1), body parts: magenta-(2), buildings: red-(3), building parts: green-(4), clothing: blue-(5), furniture: black-(6), insects: cyan-(7), kitchen items: magenta-(8), man-made objects: red-(9), tools: green-(10), vegetables: blue-(11), and vehicles: black-(12)). The computation and visualisation were made using Statistics Toolbox and Matlab.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4482269&req=5

pone.0125725.g003: MDS results for the two distance matrices on the 60 fMRI nouns.We compared the PC-score matrix of MiF-EAT (top) with the co-occurrence matrix of Mitchell et al.’s model from the Google Web 1T 5-gram collection (bottom). The natural numbers attached to the nouns represent the semantic categories (animals: cyan-(1), body parts: magenta-(2), buildings: red-(3), building parts: green-(4), clothing: blue-(5), furniture: black-(6), insects: cyan-(7), kitchen items: magenta-(8), man-made objects: red-(9), tools: green-(10), vegetables: blue-(11), and vehicles: black-(12)). The computation and visualisation were made using Statistics Toolbox and Matlab.

Mentions: It is worth noting that the semantic space underlying the 60 fMRI nouns of Mitchell et al. reflects some conceptual relationships suited to word association when using MiF applied to EAT. Fig 3 shows the results of multi-dimensional scaling (MDS) applied to the 60 × 60 MiF-EAT PC-score matrix and the 60 × 25 co-occurrence probability matrix of Mitchell et al.’s original model computed from the Google Web 1T 5-gram Collection. In MDS, each of the fMRI nouns is assigned coordinates in each dimension of the between-object distance matrix, showing the level of similarity. Some words belonging to different semantic categories become close to each other, and this closeness can be interpreted as a type of derived metonymic relationship. Contiguity (‘apartment’ and ‘bell’), target objects (‘key’ for ‘barn’ and ‘apartment’), intended or unintended uses (‘pants’ and ‘arm’, ‘hand’, ‘leg’; ‘window’ and ‘hammer’), mediated associations (‘igloo’ and ‘fridge’ through ‘cold’ or ‘icy’), and so on can be retrospectively construed as reasons for affinity (even a lexical association at the level of collocation (‘cup’–‘chisel’) might be produced ex-post facto). As for the MDS map representing the co-occurrence matrix between the nouns and the 25 basic verbs for the original Mitchell et al. model, some categories (body parts, tools) have a tendency to conglomerate at the centre, and metonymic ex-post interpretation was not as easy on the periphery as the MDS map for MiF-EAT lexical information.


Using Graph Components Derived from an Associative Concept Dictionary to Predict fMRI Neural Activation Patterns that Represent the Meaning of Nouns.

Akama H, Miyake M, Jung J, Murphy B - PLoS ONE (2015)

MDS results for the two distance matrices on the 60 fMRI nouns.We compared the PC-score matrix of MiF-EAT (top) with the co-occurrence matrix of Mitchell et al.’s model from the Google Web 1T 5-gram collection (bottom). The natural numbers attached to the nouns represent the semantic categories (animals: cyan-(1), body parts: magenta-(2), buildings: red-(3), building parts: green-(4), clothing: blue-(5), furniture: black-(6), insects: cyan-(7), kitchen items: magenta-(8), man-made objects: red-(9), tools: green-(10), vegetables: blue-(11), and vehicles: black-(12)). The computation and visualisation were made using Statistics Toolbox and Matlab.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4482269&req=5

pone.0125725.g003: MDS results for the two distance matrices on the 60 fMRI nouns.We compared the PC-score matrix of MiF-EAT (top) with the co-occurrence matrix of Mitchell et al.’s model from the Google Web 1T 5-gram collection (bottom). The natural numbers attached to the nouns represent the semantic categories (animals: cyan-(1), body parts: magenta-(2), buildings: red-(3), building parts: green-(4), clothing: blue-(5), furniture: black-(6), insects: cyan-(7), kitchen items: magenta-(8), man-made objects: red-(9), tools: green-(10), vegetables: blue-(11), and vehicles: black-(12)). The computation and visualisation were made using Statistics Toolbox and Matlab.
Mentions: It is worth noting that the semantic space underlying the 60 fMRI nouns of Mitchell et al. reflects some conceptual relationships suited to word association when using MiF applied to EAT. Fig 3 shows the results of multi-dimensional scaling (MDS) applied to the 60 × 60 MiF-EAT PC-score matrix and the 60 × 25 co-occurrence probability matrix of Mitchell et al.’s original model computed from the Google Web 1T 5-gram Collection. In MDS, each of the fMRI nouns is assigned coordinates in each dimension of the between-object distance matrix, showing the level of similarity. Some words belonging to different semantic categories become close to each other, and this closeness can be interpreted as a type of derived metonymic relationship. Contiguity (‘apartment’ and ‘bell’), target objects (‘key’ for ‘barn’ and ‘apartment’), intended or unintended uses (‘pants’ and ‘arm’, ‘hand’, ‘leg’; ‘window’ and ‘hammer’), mediated associations (‘igloo’ and ‘fridge’ through ‘cold’ or ‘icy’), and so on can be retrospectively construed as reasons for affinity (even a lexical association at the level of collocation (‘cup’–‘chisel’) might be produced ex-post facto). As for the MDS map representing the co-occurrence matrix between the nouns and the 25 basic verbs for the original Mitchell et al. model, some categories (body parts, tools) have a tendency to conglomerate at the centre, and metonymic ex-post interpretation was not as easy on the periphery as the MDS map for MiF-EAT lexical information.

Bottom Line: We measure the effectiveness of graph-based coefficients through the application of linguistic graph information for a neural activity recorded during conceptual processing in the human brain.Furthermore, correlating the voxel information with the MiF-based principal components, a new computational neurolinguistics model with a network connectivity paradigm is created.This allows two dimensions of context space to be incorporated with both semantic and neural distributional representations.

View Article: PubMed Central - PubMed

Affiliation: Graduate School of Decision Science and Technology, Tokyo Institute of Technology, Tokyo, Japan.

ABSTRACT
In this study, we introduce an original distance definition for graphs, called the Markov-inverse-F measure (MiF). This measure enables the integration of classical graph theory indices with new knowledge pertaining to structural feature extraction from semantic networks. MiF improves the conventional Jaccard and/or Simpson indices, and reconciles both the geodesic information (random walk) and co-occurrence adjustment (degree balance and distribution). We measure the effectiveness of graph-based coefficients through the application of linguistic graph information for a neural activity recorded during conceptual processing in the human brain. Specifically, the MiF distance is computed between each of the nouns used in a previous neural experiment and each of the in-between words in a subgraph derived from the Edinburgh Word Association Thesaurus of English. From the MiF-based information matrix, a machine learning model can accurately obtain a scalar parameter that specifies the degree to which each voxel in (the MRI image of) the brain is activated by each word or each principal component of the intermediate semantic features. Furthermore, correlating the voxel information with the MiF-based principal components, a new computational neurolinguistics model with a network connectivity paradigm is created. This allows two dimensions of context space to be incorporated with both semantic and neural distributional representations.

No MeSH data available.