Search

Search Results

• ALGORITHM FOR FILTERING "HINT INJECTIONS" WHEN USING SPATIAL INFORMATION

  S.L. Belyakov , L.А. Izrailev , О.N. Pokusaev

  32-45

  2026-07-07

  Abstract ▼

  The integration of large language models (LLM) into geographic information systems (GIS) opens up new opportunities for spatial analysis, but it is accompanied by specific vulnerabilities such as "hint injection" (prompt injection). Such attacks allow attackers to bypass LLM security mechanisms, manipulate issuance, gain access to confidential information, and violate data integrity. Using space allows you to access an object not directly, but through its spatial relationships with other objects. Existing keyword or template filtering methods do not provide reliable protection due to the constant emergence of new attack scenarios. This determines the relevance of developing adaptive, self-learning algorithms for filtering queries for industrial injections to large language models. The aim of the study is to develop an algorithm for filtering prompt injections for LLM, based on the Case-Based Reasoning (CBR) method. The paper proposes an algorithm for comparing LLM queries with a database of previously known promt injections. The experiment showed that as the database of use cases accumulates, the accuracy of detecting prompt injections increases from 42% to 83%. At the same time, the processing time for a single request increases slightly (from 0.18 to 0.19 seconds with a 23% increase in the database). Approaches to generalizing the precedent base and introspection of the precedent base were also proposed. The proposed algorithm makes it possible to increase the security of LLM-interface systems against prompt injections due to adaptivity and self-learning. The practical significance lies in the possibility of implementing the developed filter into information systems to prevent leaks and manipulation of spatial data. Further research is related to the development of methods for automatic generalization of use cases and the integration of additional contextual analyzers.

• GEOINFORMATION MODELS OF EMERGENCY SITUATIONS WITH SPATIAL GENERALIZATIONS

  S. L. Belyakov, L. А. Izrailev

  2025-01-30

  Abstract ▼

  The main problem of decision making in emergency situations is the reliability of these decisions. Emergency
  situations by virtue of its unpredictable and dynamic nature often have incomplete and inaccurate information.
  The use of accumulated experience allows to find reliable solutions based on known precedents of
  emergency situations. Geographic information systems (GIS) can act as a tool for accumulating experience and
  generating solutions based on it. The cartographic basis of GIS allows analyzing emergency situations, taking
  into account their spatial and temporal characteristics. However, the cartographic representation of precedents
  with adopted solutions describes them too narrowly. There is no idea what properties of the situation are significant
  and whether the precedent solution can be applied in other circumstances. The use of known images and
  their admissible transformations, created on the basis of expert knowledge, can solve this problem. The image
  generalizes a set of similar precedents. The purpose of such generalization is to expand the area of application
  of information from private observations by determining the boundaries of permissible transformations. However,
  the need to attract experts for their creation is a difficult task, since each situation is unique in its own way.
  No less problematic is the transfer of experience from one spatial and temporal domain to another. In this paper
  we consider an approach to automatic image generation. We propose a method of creating a geoinformation
  model of emergency situations, which includes the generalization of precedents on a common location. This
  approach is aimed at improving the reliability of prediction of emergency situations. An experiment was conducted
  to synthesize images based on precedents of road accidents and evaluate their effectiveness compared to
  individual precedents. The use of the developed method of automatic data processing to create images is relevant,
  as it significantly reduces the cost of knowledge acquisition. The use of spatial generalizations also eliminates
  the need for expert knowledge, since the formation of precedent sets is performed by analyzing their geographical
  location.