Because of it is applicability in various field, multi-instance learning or multi-instance problem becoming more popular in machine learning research field. Different from supervised learning, multi-instance learning related to the problem of classifying an unknown bag into positive or negative label such that labels of instances of bags are ambiguous. This paper uses and study three different k-nearest neighbor algorithm namely Bayesian -kNN, citation -kNN and Bayesian Citation -kNN algorithm for solving multi-instance problem. Similarity between two bags is measured using Hausdroff distance. To overcome the problem of false positive instances constructive covering algorithm used. Also the problem definition, learning algorithm and experimental data sets related to multi-instance learning framework are briefly reviewed in this paper.