Most web and mobile applications are based on searching the location-based objects called spatial objects. In spatial database systems, searching such objects is a challenging task since it deals with geo-spatial capabilities. Sometimes, the spatial queries are associated with text information in order to obtain the most relevant answers nearest to the given location. Such queries are called spatial textual query. Conventional spatial indexes and text indexes are not suitable for resolving such queries. Since these indexes use various approaches to perform searching, they can cause performance degradation. Effective processing of the query mainly depends on the index structure, searching algorithms, and location-based ranking. This chapter reviews the different hybrid index structures and search mechanisms to extract the spatial objects, the different ranking model it supports, and the performance characteristics.
TopIntroduction
Spatial query is the special type of database query supported by the geospatial databases or spatial databases. Nowadays, most of the research areas are focused on web applications and mobile based applications. Often these applications are in need to search the answer for the queries and find out the most relevant answer to the query. The query may result in finding the spatial objects shown in Figure 1 that are nearest to the given locations called location based objects.
Sometimes the spatial query can also be enriched with some textual information called keywords as shown in Table 1. Each object has associated text information and the number of times its occurrence within the document. This type of query is referred as Spatial-Textual query (ST). For example, the query may be in the form “Find the list of hotels within 2km with the facility of pool”. ST query can be resolved by resolving two predicates called spatial predicates and string predicates. In order to resolve these predicates, different types of index structures which optimizes the search space are constructed
Table 1. Objects associated with text information
| Object | Text information Associated |
| O1 | (AC,5) (Box,5) (Expensive,3) |
| O2 | (Café,5) (Box,4) (Expensive,2) |
| O3 | (AC,3) (Balcony, 3) (Expensive,1) |
| O4 | (Box,4) (Balcony,2) (Expensive,2) |
| O5 | (AC, 3) (Online reservation,4) |
| O6 | (Cafe,3)(Online reservation,2) |
Index structure for spatial predicates falls under three categories. They are tree based, grid based and Quad tree based Index structures. Inverted file index, bitmap index and signature file are used to resolve the string predicates. Initially, two approaches such as Geospatial approach and textual approach are used to resolve the ST query. In Geospatial approach, the spatial predicates are resolved first and then the textual information is searched. The textual approach, first output the query answer that satisfies the textual information, and then the results are checked against the spatial coordinates. In order to reduce the search space, a new hybrid index structure which combines both the approach is required. Each node is augmented with both textual as well as the spatial information. It is necessary to consider both the spatial indexing as well as the text indexing to retrieve the spatial object. This paper discusses the different indexing structure and search mechanisms that support both spatial and text information.
TopIndex Structure Classification
Queries with spatial condition can be answered by a tree based index like R tree and its variants (Bechmann, Kriegel, Schneider & Seeger, 1990; Guttman, 1984), KD tree, Quad tree, etc. R tree is used as the most effective basic index shown in Figure 2 for representing the spatial condition.