Conflation



Conflation occurs when the identities of two or more individuals, concepts, or places, sharing some characteristics of one another, become confused until there seems to be only a single identity — the differences appear to become lost.

In logic, the practice of treating two distinct concepts as if they were one often produces error or misunderstanding — but not always — as a fusion of distinct subjects tends to obscure analysis of relationships which are emphasized by contrasts.

Conflation as a GIS-technology
In GIS, conflation is defined as the process of combining geographic information from overlapping sources so as to retain accurate data, minimize redundancy, and reconcile data conflicts. (quoted [Longley et al. 2001], see [Seth, Samal 2008, p129])

Horizontal conflation refers to the matching of features and attributes in adjacent GIS sources for the purpose of eliminating positional and attribute discrepancies in the common area of the two sources.

Vertical conflation solves a similar problem for GIS sources with overlapping coverage.

As features are the basic entities in a GIS, the special case of feature conflation has received much attention in the published research.

The data used for conflation are point, line, and area features and their attributes. ... With conflation ... discrepancies can be minimized, thus improving the overall accuracy of the data sources.

The goal of conflation is to combine the best-quality elements of both datasets to create a composite dataset that is better than either of them. The consolidated dataset can then provide more information than can be gathered from any single dataset. [Chen, Knoblock, 2008, p133ff]

Conflation is a combination of two digital maps to produce a third map file which is "better" than each of the component source maps (Lynch and Saalfeld, 1985).

Vector-to-raster and raster-to-raster conflation are also referenced as image registration. "Image Registration Methods: A Survey" (Barbara Zitova, 2003) present recent progress about image registration.

Historical Background
Automated vector and vector conflation was first proposed by Saalfeld [1], and the initial focus of conflation was using geometrical similarities between spatial attributes (e.g., location, shape, etc.) to eliminate the spatial inconsistency between two overlapping vector maps. In particular, in [1], Saalfeld discussed mathematical theories to support the automatic process. From then, various vector-to-vector conflation techniques have been proposed [2,3] and many GIS systems have been implemented to achieve the alignments of geospatial datasets. More recently, with the proliferation of attributed vector data, attribute information (i. e., non-spatial information) has become another prominent feature used in the conflation systems, such as ESEA MapMerger and the system developed by Cobb et al. [4]. Most of the approaches mentioned above focus on vector-to-vector conflation by adapting different techniques to perform the matching.

(...) The framework of conflation process can be generalized into the following steps:

1) Feature matching: Find a set of conjugate point pairs, "termed control point pairs," in two datasets,

2) Match checking: Filter inaccurate control point pairs from the set of control point pairs for quality control, and

3) Spatial attribute alignment: Use the accurate control points to align the rest of the geospatial objects (e. g., points or lines) in both datasets by using space partitioning techniques (e. g., triangulation) and geometric interpolation techniques.

During the late 1980s, Saalfeld [1] initialized the study to automate the conflation process. He provided a broad mathematical context for conflation theory. In addition, he proposed an iterative conflation paradigm based on the above-mentioned conflation framework by repeating the matching and alignment, until no further new matches are identified. In particular, he investigated the techniques to automatically construct the influence regions around the control points to reposition other features into alignment by appropriate local interpolation (i. e., to automate the third step in the above-mentioned conflation framework).

The conclusion of Saalfeld’s work is that Delaunay triangulation is an effective strategy to partition the domain space into triangles (influence regions) to define local adjustments (see the example in Fig. 4).

A Delaunay triangulation is a triangulation of the point set with the property that no point falls in the interior of the circumcircle of any triangle (the circle passing through the three triangle vertices). (...) It maximizes the minimum angle of all the angles in the triangulation, thus avoiding elongated, acute-angled triangles. (...)

Vector-to-vector conflation
There have been a number of efforts to automatically or semi-automatically accomplish vector-to-vector conflation. Most of the existing vector-vector conflation algorithms focus on road vector data. These approaches are different, because of the different methods used to locate the counterpart elements from both vector datasets.

The major approaches include: – Matching vector data based on the similarities of geometric information (such as nodes and lines) [1,2,3]. – Matching attribute-annotated vector data based on the similarities of vector shapes as well as the semantic similarities of vector attributes [4]. – Matching vector data with unknown coordinates based on the feature point (e. g., the road intersection) distributions [13].

[1] Saalfeld, A.: Conflation: Automated Map Compilation. Int. J. Geogr. Inf. Sci. 2(3), 217–228 (1988)

[2] Walter, V., Fritsch, D.: Matching Spatial Data Sets: a Statistical Approach. Int. J. Geogr. Inf. Sci. 5(1), 445–473 (1999)

[3] Ware, J.M., Jones, C.B.: Matching and Aligning Features in Overlayed Coverages. In: Proceedings of the 6th ACM Symposium on Geographic Information Systems. Washingnton, D.C. (1998)

[4] Cobb, M., Chung, M.J., Miller, V., Foley, H. III, Petry, F.E., Shaw, K.B.: A Rule-Based Approach for the Conflation of Attributed Vector Data. GeoInformatica 2(1), 7–35 (1998)

[13] Chen, C.-C., Shahabi, C., Knoblock, C.A., Kolahdouzan, M.: Automatically and Efficiently Matching Road Networks with Spatial Attributes in Unknown Geometry Systems. In: Proceedings of the Third Workshop on Spatio-temporal Database Management (co-located with VLDB2006). Seoul, Korea (2006)

Taxonomic conflation
In scientific taxonomies, a conflative term is always a polyseme.