INVESTIGATING THE SEMANTIC SIMILARITY EFFECT ON DELAYED FREE RECALL USING WORD EMBEDDINGS

Download
2025-1-10
Büyükyaprak, Burak
Episodic memory is a type of long-term memory that encodes and retrieves personal experiences associated with their context. Previous episodic memory studies showed that the context or pre-existing knowledge about retrieved information may influence the performance of memory tasks. So, it becomes crucial to study the semantic proximity effect by comparing memory task performance with different levels of semantic relatedness. In natural language processing studies, semantic relations can be successfully represented by learning word vectors in a large text corpus using neural networks. The study aimed to investigate the impact of semantic factors on delayed free recall task by creating word lists that included semantically related and unrelated word lists obtained through neural networks and show how semantic and temporal proximity influence recall performance. fastText and Word2Vec were used to obtain Turkish word representations and organize words according to their semantic relatedness, and human raters validated words in the word lists. Later, to investigate how semantic relatedness affects recall dynamics, four conditions were compared (fastText-related, fastText-unrelated, Word2Vec-related, Word2Vec-unrelated). Results showed that a significant positive correlation between cosine similarity values and human judgments, and semantic and temporal proximity effects influenced the recall probability. In addition, different levels of semantic relatedness and choice of word embeddings played a role in the likelihood of recall and accuracy. Therefore, this thesis suggests that neural networks can represent and manipulate semantic relations in memory studies and that semantic and temporal proximity influence recall dynamics at varying levels of semantic relatedness.
Citation Formats
B. Büyükyaprak, “INVESTIGATING THE SEMANTIC SIMILARITY EFFECT ON DELAYED FREE RECALL USING WORD EMBEDDINGS,” M.S. - Master of Science, Middle East Technical University, 2025.