Semantic clustering on common list-learning tasks: a systematic review of the state of the literature and recommendations for future directions

INTRODUCTION

On some list-learning tasks, such as the California Verbal Learning Test (CVLT) or Hopkins Verbal Learning Test (HVLT), examinees have the opportunity to group words based on semantically related categories (i.e., semantic clustering). Semantic clustering (SC) is often considered the most efficient organizational strategy and adopting SC is presumed to improve learning and memory. In addition, SC is conceptualized as reflecting higher-order executive functioning skills. Although SC measures have intuitive appeal, to date, there are no comprehensive reviews of the SC literature base that summarize its psychometric utility. In this systematic review, we synthesize the literature to judge the validity of SC scores.

METHOD

We conducted a systematic literature search for empirical articles reporting SC from the CVLT and HVLT. We qualitatively described the relationship of SC with other list-learning and cognitive test scores and clinical diagnoses, contrasting SC with serial clustering and total learning scores when possible.

RESULTS

SC was inversely correlated with serial clustering. Higher SC was strongly associated with better learning and memory performances. When compared with cognitive tests, SC tended to have the strongest relationships with other memory measures and modest relationships with tests of executive functioning. SC had negligible to small relationships with most other cognitive domains. Traditional memory scores yielded stronger relationships to cognitive test performances than did SC. SC across clinical groups varied widely, but clinical groups tended to use SC less often than healthy comparison groups.

CONCLUSION

Our comprehensive review of the literature revealed that SC is strongly related to measures of learning and memory on the CVLT and HVLT and is correlated with a wide range of cognitive functions. SC has been understudied in relevant populations and additional research is needed to test the degree to which it adds incremental validity beyond traditional measures of learning and memory.

Adults with overweight or obesity use less efficient memory strategies compared to adults with healthy weight on a verbal list learning task modified with food words

Several studies suggest poorer episodic memory among adults with overweight (OW) relative to those with healthy weight (HW); however, few have used food stimuli. To understand the salience of food-related items when assessing memory, we adapted an episodic memory task, by replacing some non-food words with snack foods. Participants were 96 weight-loss seeking adults with OW compared to 48 adults with HW from the community matched on age, gender, ethnicity, and education. Overall memory ability was similar, although a trend showed the adults with HW performed better than adults with OW on immediate recall (d = 0.32, p = 0.07). However, there were clear differences in the use of learning strategies. Adults with HW utilized sematic clustering more effectively than adults with OW during all test phases (ds = 0.44-0.62; ps ≤ 0.01). Adults with HW also utilized serial clustering more effectively (d = 0.51; p < 0.01). Adults with HW showed better semantic clustering for both food and non-food words during immediate and short delay recall (ds = 0.42-0.78; ps ≤ 0.01) but semantic clustering was only better for the non-food category at long delay (d = 0.55; p < 0.01). These results show that adults with OW utilized less efficient learning strategies throughout the task and food-related content may impact learning. Clinically, these findings may suggest that weight-loss treatments should consider incorporating the teaching of learning and memory strategies to help increase utilization of new skills.

WEClustering: word embeddings based text clustering technique for large datasets

A massive amount of textual data now exists in digital repositories in the form of research articles, news articles, reviews, Wikipedia articles, and books, etc. Text clustering is a fundamental data mining technique to perform categorization, topic extraction, and information retrieval. Textual datasets, especially which contain a large number of documents are sparse and have high dimensionality. Hence, traditional clustering techniques such as K-means, Agglomerative clustering, and DBSCAN cannot perform well. In this paper, a clustering technique especially suitable to large text datasets is proposed that overcome these limitations. The proposed technique is based on word embeddings derived from a recent deep learning model named “Bidirectional Encoders Representations using Transformers”. The proposed technique is named as WEClustering. The proposed technique deals with the problem of high dimensionality in an effective manner, hence, more accurate clusters are formed. The technique is validated on several datasets of varying sizes and its performance is compared with other widely used and state of the art clustering techniques. The experimental comparison shows that the proposed clustering technique gives a significant improvement over other techniques as measured by metrics such Purity and Adjusted Rand Index.

BioPREP: Deep learning-based predicate classification with SemMedDB

When it comes to inferring relations between entities in biomedical texts, Relation Extraction (RE) has become key to biomedical information extraction. Although previous studies focused on using rule-based and machine learning-based approaches, these methods lacked efficiency in terms of the demanding amount of feature processing while resulting in relatively low accuracy. Some existing biomedical relation extraction tools are based on neural networks. Nonetheless, they rarely analyze possible causes of the difference in accuracy among predicates. Also, there have not been enough biomedical datasets that were structured for predicate classification. With these regards, we set our research goals as follows: constructing a large-scale training dataset, namely Biomedical Predicate Relation-extraction with Entity-filtering by PKDE4J (BioPREP), based on SemMedDB then using PKDE4J as an entity-filtering tool, evaluating the performances of each neural network-based algorithms on the structured dataset. We then analyzed our model's performance in-depth by grouping predicates into semantic clusters. Based on comprehensive experimental outcomes, the experiments showed that the BioBERT-based model outperformed other models for predicate classification. The suggested model achieved an f1-score of 0.846 when BioBERT was loaded as the pre-trained model and 0.840 when SciBERT weights were loaded. Moreover, the semantic cluster analysis showed that sentences containing key phrases were classified better, such as comparison verb + 'than'.

Investigating the trends in arctic research: The increasing role of social sciences and humanities

The Arctic Region experienced a series of significant changes due to shifting climate conditions, resulting in multiple opportunities and challenges for international actors, and encouraging both Arctic and non-Arctic states to promote their own national interests. Hence, the region has become a global priority, and a focus of scientific studies across the Natural Sciences, and Social Sciences and Humanities (SSH) disciplines. This study systematically analyses the literature on the Arctic Region, conducting a multidimensional bibliometric analysis and content analysis on the basis of semantic clustering. The purpose of the analysis is to determine future Arctic-related research themes. The study follows a three-level research framework. The first level of the analysis highlights a disciplinary shift in the Arctic literature from Natural Sciences towards Social Sciences and Humanities, particularly, focusing on the environment, technology, political and energy-related issues. The second level identifies 9 research themes which are validated in the third level. The third level reveals the most prominent terms and prioritized research areas in the Arctic literature, namely, Governance, Security Issues, Economic Factors, Legal Issues, Energy and Natural Resources, Logistics, Climate Change and Environment, Technology, and Socio-cultural and Ethnic Issues.

Medial Temporal Lobe Atrophy is Related to Learning Strategy Changes in Amnestic Mild Cognitive Impairment

OBJECTIVE

Deficits in the semantic learning strategy were observed in subjects with amnestic mild cognitive impairment (aMCI) in our previous study. In the present study, we explored the contributions of executive function and brain structure changes to the decline in the semantic learning strategy in aMCI.

METHODS

A neuropsychological battery was used to test memory and executive function in 96 aMCI subjects and 90 age- and gender-matched healthy controls (HCs). The semantic clustering ratio on the verbal learning test was calculated to evaluate learning strategy. Medial temporal lobe atrophy (MTA) and white matter hyperintensities (WMH) were measured on MRI with the MTA and Fazekas visual rating scales, respectively.

RESULTS

Compared to HCs, aMCI subjects had poorer performance in terms of memory, executive function, and the semantic clustering ratio (P < .001). In aMCI subjects, no significant correlation between learning strategy and executive function was observed. aMCI subjects with obvious MTA demonstrated a lower semantic clustering ratio than those without MTA (P < .001). There was no significant difference in the learning strategies between subjects with high-grade WMH and subjects with low-grade WMH.

CONCLUSION

aMCI subjects showed obvious impairment in the semantic learning strategy, which was attributable to MTA but independent of executive dysfunction and subcortical WMH. These findings need to be further validated in large cohorts with biomarkers identified using volumetric brain measurements. (JINS, 2019, 25, 706-717).

Executive function predictors of delayed memory deficits after mild traumatic brain injury

Delayed memory deficits are common for patients with mild traumatic brain injury (mTBI), according to a recent systematic review of meta-analyses (Karr et al., 2014). However, there has been little work to identify different cognitive processes that may be underpinning these delayed memory deficits for mTBI. Frontal cortex is important for delayed memory, and is implicated in the pathophysiology of mTBI; moreover, frontal lobes are typically considered the locus of executive abilities. To further explore these relationships, we sought to partly explain delayed memory deficits after mTBI by examining behavioral indicators of executive function. Results showed that sub-acute as well as chronic mTBI patients performed worse than controls on the delayed memory trial of the Hopkins Verbal Learning Test-Revised (Brandt & Benedict, 2001), recalling approximately 18% and 15% fewer words, respectively. Furthermore, worse delayed memory performance was associated with less use of the cognitive strategy of semantic clustering, and with lower scores for the executive function composite from a standardized neuropsychological battery (NIH EXAMINER; Kramer et al., 2014). In contrast, serial clustering, a memory organizational strategy thought to be less dependent on executive function, did not show strong relationships to clinical status or delayed memory performance. This exploratory work suggests novel hypotheses to be tested in future, confirmatory studies, including that general executive functions and/or semantic clustering will mediate delayed memory deficits following mTBI.

Organizational Learning Strategies and Verbal Memory Deficits in Bipolar Disorder

BACKGROUND

Verbal memory (VM) impairment is prominent in bipolar disorder (BD) and is linked to functional outcomes. However, the intricacies of VM impairment have not yet been studied in a large sample of BD patients. Moreover, some have proposed VM deficits that may be mediated by organizational strategies, such as semantic or serial clustering. Thus, the exact nature of VM break-down in BD patients is not well understood, limiting remediation efforts. We investigated the intricacies of VM deficits in BD patients versus healthy controls (HCs) and examined whether verbal learning differences were mediated by use of clustering strategies.

METHODS

The California Verbal Learning Test (CVLT) was administered to 113 affectively stable BD patients and 106 HCs. We compared diagnostic groups on all CVLT indices and investigated whether group differences in verbal learning were mediated by clustering strategies.

RESULTS

Although BD patients showed significantly poorer attention, learning, and memory, these indices were only mildly impaired. However, BD patients evidenced poorer use of effective learning strategies and lower recall consistency, with these indices falling in the moderately impaired range. Moreover, relative reliance on semantic clustering fully mediated the relationship between diagnostic category and verbal learning, while reliance on serial clustering partially mediated this relationship.

CONCLUSIONS

VM deficits in affectively stable bipolar patients were widespread but were generally mildly impaired. However, patients displayed inadequate use of organizational strategies with clear separation from HCs on semantic and serial clustering. Remediation efforts may benefit from education about mnemonic devices or "chunking" techniques to attenuate VM deficits in BD. (JINS, 2017, 23, 358-366).

Language networks associated with computerized semantic indices

Tests of generative semantic verbal fluency are widely used to study organization and representation of concepts in the human brain. Previous studies demonstrated that clustering and switching behavior during verbal fluency tasks is supported by multiple brain mechanisms associated with semantic memory and executive control. Previous work relied on manual assessments of semantic relatedness between words and grouping of words into semantic clusters. We investigated a computational linguistic approach to measuring the strength of semantic relatedness between words based on latent semantic analysis of word co-occurrences in a subset of a large online encyclopedia. We computed semantic clustering indices and compared them to brain network connectivity measures obtained with task-free fMRI in a sample consisting of healthy participants and those differentially affected by cognitive impairment. We found that semantic clustering indices were associated with brain network connectivity in distinct areas including fronto-temporal, fronto-parietal and fusiform gyrus regions. This study shows that computerized semantic indices complement traditional assessments of verbal fluency to provide a more complete account of the relationship between brain and verbal behavior involved organization and retrieval of lexical information from memory.