Non-separability Effects in Cognitive Semantic Retrieving

This paper discusses a Bell test analogue known in quantum physics, which allows determining the presence of non-separability features by using semantic search of information and document ranking for articles in Russian. The model of Bell test in semantics is based on hyperspace analogue to language (HAL) algorithm provides to obtain vector representation of words (in Hilbert space) using the dictionary index and considering the word order. We show the existence of certain quantum-like correlations between two words of the user’s query; these correlations cannot be taken into account in the classical probabilistic description. We predict that the contextuality revealed can be regarded as human cognitive level both while writing of certain texts and queries to them.