A Novel Interdisciplinarity Model Towards Inter-domain Information Pairing

Abstract

This study introduces an interdisciplinary prediction frameworkas part of a novel approach that integrates the Inventive DesignMethod (IDM), Topic Modeling, and Generative AI to foster innovationacross academic fields. Identifying interdisciplinary connections is essentialfor solving complex, multi-domain problems. Our research uses asupervised machine learning classifier to identify interdisciplinary documentswithin the Semantic Scholar corpus, extracting latent insights. TheText Convolutional Neural Network model performed best, achieving anF1 score of 0.80. We find that approximately 25% of human knowledge isinterdisciplinary. This framework helps create comprehensive knowledgemaps across multiple domains, promoting innovation through effectivecross-domain knowledge transfer.

BibTeX

@inproceedings{douard:hal-04922408,
  TITLE = {{A Novel Interdisciplinarity Model Towards Inter-domain Information Pairing}},
  AUTHOR = {Douard, Nicolas and Samet, Ahmed and Giakos, George and Cavallucci, Denis},
  URL = {https://hal.science/hal-04922408},
  NOTE = {Part 3: Interdisciplinary and Cognitive Approaches in TRIZ},
  BOOKTITLE = {{IFIP Advances in Information and Communication Technology}},
  ADDRESS = {Cluj-Napoca, Romania},
  EDITOR = {Denis Cavallucci and Stelian Brad and Pavel Livotov},
  PUBLISHER = {{Springer Nature Switzerland}},
  VOLUME = {AICT-736},
  YEAR = {2024},
  MONTH = Nov,
  KEYWORDS = {language models ; topic models ; literature analysis ; biomimicry ; cross-domain transfer ; Interdisciplinary mapping},
  PDF = {https://hal.science/hal-04922408v1/file/islandora_174002.pdf},
  HAL_ID = {hal-04922408},
  HAL_VERSION = {v1},
}