Mining Spatial Co-location Patterns With Dominant Features Based On Fuzzy Neighborhood Relationships

In recent years,with the rapid development and application of spatial data acquisition tools such as global positioning system,information system and mobile devices,a large amount of spatial data containing location information has been produced.Co-location pattern mining is to discover subsets of spatial features that frequently appear together in a space from massive spatial data.The traditional frequent juxtaposition model excavates a large number of patterns,and the patterns only express the coexistence information of spatial characteristics,which is lack of guidance in practical application.In order to solve this problem,the unequal relationship of features in frequent juxtaposition patterns was found by comparing the interaction between features and patterns,and then the frequent juxtaposition patterns with dominant features were excavated to improve the practicability and decision ability of the patterns.Although containing leading character juxtaposition pattern mining results have more abundant semantic also more practical,but existing methods exist the following problems:first,the dominant characteristics of the existing juxtaposition pattern mining method without considering the space between the instances of adjacent relation is a fuzzy concept,did not take into account the neighboring distance size influence on adjacent level,has caused the relationship between some neighboring missing or incorrect.Secondly,the existing methods of juxtaposition pattern mining with dominant features fail to consider that the meaning of dominant is also a fuzzy concept,and adopt an either-or method,namely the 0/1 measure,when measuring instance contribution,pattern contribution and pattern influence.Although the juxtaposition pattern mining method with dominant features greatly improves the quality and guiding function of the traditional frequent juxtaposition pattern mining method,the results still have problems due to the lack of consideration of the fuzziness of corresponding concepts.In this paper,the fuzzy set theory is introduced into the juxtaposition pattern mining with dominant features,considering that both proximity relation and dominant relation are fuzzy concepts.Firstly,the fuzzy proximity relationship between spatial instances is defined based on the fuzzy set theory,and the proximity membership function is given,and the proximity degree is calculated according to the fuzzy membership function.Secondly,the concepts of contribution degree of instance fuzzy row instance and pattern contribution degree of instance are proposed,which are used as the calculation criteria of feature participation rate and pattern participation rate,and meet the downward closure property.At the same time,in order to compare the influence relationship between features in the juxtaposition model,the concepts of fuzzy loss rate,fuzzy influence degree and fuzzy influence ratio of features were defined,and the dominant features in the model were identified.Finally,the basic algorithm DFMAFPR and the optimal algorithm DFMAFPR-Improved for mining spatial collocation patterns with dominant features based on fuzzy neighborhood relations are proposed,and the correctness and effectiveness of the proposed algorithm are verified on synthetic and real datasets.