Computational Tool for Automatic Term Extraction - ATEM

Adapting TBXTools to automatic terminology extraction with BERT

Efficient Visualization of Knowledge Graphs Based on Next-Generation Language Models

Introduction . This paper examines a methodology for the automated extraction and graphical representation of knowledge from unstructured texts using modern language models. Such methods are becoming increasingly relevant because of the growing need to structure information and identify semantic relations that are difficult to capture manually. Materials and Methods . The proposed approach combines locally deployed language models with specialized relationextraction tools. Local deployment enables data to be processed in a secure environment without reliance on external services. The methodology includes text preprocessing, entity and relation extraction, structuring, and visualization of the resulting knowledge graphs. Results . Experimental testing on a corpus of Russian-language scientific articles demonstrated that the approach is applicable both to technical descriptions and to texts containing more abstract concepts. The developed web interface supports interactive visualization and comparative analysis of graphs constructed by different models, thereby improving the interpretability of the results. The approach is robust to textual noise and is applicable to scientific, technical, and regulatory tasks. Discussion . The results show that the proposed methodology is not limited to a single algorithm and permits the combination of direct extraction, specialized models, multi-stage pipelines, and OWL ontologies. The quality of the resulting graphs depends substantially on the structure of the source text, preprocessing accuracy, and the selected postprocessing procedures; interactive visualization facilitates comparison of outputs generated by different models and supports the interpretation of semantic relations. Conclusions . The proposed approach can be applied to the analysis of scientific, technical, and regulatory texts in a secure local environment. It is a natural continuation of the authors’ previous research on semantic-associative data analysis and synthesis and the associative-ontological approach. Further development should focus on ensemble schemes, logical validation, semantic inference, and integration with formal ontologies, thereby extending its applicability to information retrieval, research support, and complex-system modelling.

Assessment of Term Extraction Methodology Applied to Terminology Work. A Case Study of Catalan Legal Domain

Structure then Query: Enabling Precise Analytical Queries over Unstructured Documents

Experiments on three real-world datasets demonstrate that AnnoIndex consistently outperforms state-of-the-art baselines, achieving the highest average F1 score while maintaining robust performance on complex multi-hop join and progressive reasoning queries.

Automated Summarization Tool

The design realization and evaluation of an Automated Summarization Tool (AST) is presented which is a document intelligence platform based on google gemini 2.5 flash that outperforms the strongest fine-tuned transformer baselines (PEGASUS, BART) by ~14 points and is clearly ahead of BERTSUM-ext (a strong transformer baseline), Pointer-Generator Network, TextRank.

Ontology-Based Semantic Normalization of Resumes for Classification

Rather than scaling performance uniformly across the entire evaluation suite, the ontology layer acts as a targeted traceability and semantic refinement filter that contributes information beyond filtered-profile selection alone and produces a metric-dependent change in classifier behaviour at the validation-selected threshold.