Maintaining Relational Consistency in a Graph-Based Place Database Hao Chen Department of Infrastructure Engineering, Geomatics The University of Melbourne Email: hchen@student.unimelb.edu.au
Introduction Natural language (NL) descriptions & shared spatial knowledge places
Place Description Place The referent of locative noun phrase i.e. Library, Park, Melbourne Place description Locatum, relatum, relation i.e. The library is near the park Spatial Triplet Model of place description i.e. <library, near, park> http://telluswhere.net/
Graph-Based Place Database The library is near the park label {name: library} A B label {name: park} NL place Description Property Graph (2 nodes and 1 directed edge) NL parsing (Kim, Vasardani et al. 2014) <library, near, park> Triplet <locatum, relation, relatum> www.insidefacebook. com Graph Database - Efficient (avoid joining tables) - Suitable for modelling rich-connected data
Problem & Objective Problem Relational inconsistent i.e. <A, north_of, B>, <B, north_of, C> and <C, north_of, A> - logical contradictions North A B C East Objectives Develop relational consistency rules Major Tools - Qualitative Spatial Reasoning (QSR) (Cohn and Hazarika 2001, Dylla, Frommberger et al. 2006)
Relational Consistency (1/3) Relation Families Cardinal direction relations: Frank s neutral zone model (Frank 1991) North, Northeast, East, Southeast, South, Southwest, West, Northwest Topological relations 4-intersection model (Egenhofer and Herring 1990) Disjoint, Meet, Equal, Inside, CoveredBy, Contains, Covers, Overlap Qualitative distance relations Multi-step systems (Frank 1992) Near, Middle, Far Not applied in this research: Relative direction relations In front of, Right of, Behind, Left of
Relational Consistency (2/3) Relation Composition Composition operation: - Between two relations r A,B and r B,C - Returns a set of possible relations between A and C - r A,B r B,C = S A,C Composition table Example: cardinal direction family (Frank 1991)
Relational Consistency (2/3) Relation Composition (con.) Example
Relational Consistency (3/3) Checking Mechanism Step 1: - For a new input <A, r, B> - Find all paths between A and B Step 2: - Composite r1 r2 r3 = S A,B - If r S A,B, then <A, r, B> is consistent Step 3: Check for all paths
Implementation (1/3) - Overview Dataset 42 campus descriptions of the University of Melbourne s Parkville By volunteer graduate students. Programming Neo4j + Python-embedded Experiments Test for all triplets + made-up triplets Flag inconsistency
Implementation (2/3) Checking Procedure triplets node already exist? no create node yes checking procedure cardinal direction relationship approximate qualitative distance relationship topological relationship other rule 1 rule rule 2 rule 3 rule rule 4 rule 5 rule rule 6 reject flag no consitent? yes create L-r->R
Implementation (3/3) Place Database
Spatial Queries
Discussion & Future Work The database is robust for preserving relational consistency It only flags inconsistency, not decide which one is truer Future works include: - Semantic consistency of nodes - Including relative direction relations - Integrate reasoning of different families - Relations lost their context
References Cohn, A. G. and S. M. Hazarika (2001). "Qualitative spatial representation and reasoning: An overview." Fundamenta informaticae 46(1): 1-29. Dylla, F., L. Frommberger, J. O. Wallgrün and D. Wolter (2006). SparQ: A toolbox for qualitative spatial representation and reasoning. Qualitative Constraint Calculi: Application and Integration, Workshop at KI. Egenhofer, M. J. and J. Herring (1990). "Categorizing binary topological relations between regions, lines, and points in geographic databases." 9: 94-91. Egenhofer, M. J. and R. D. Franzosa (1991). "Point-set topological spatial relations." International Journal of Geographical Information System 5(2): 161-174. Frank, A. U. (1991). Qualitative spatial reasoning with cardinal directions. Österreichische Artificial-Intelligence-Tagung/Seventh Austrian Conference on Artificial Intelligence, Springer. Frank, A. U. (1992). "Qualitative spatial reasoning about distances and directions in geographic space." Journal of Visual Languages & Computing 3(4): 343-371. Kim, J., M. Vasardani and S. Winter (2014). "From Descriptions to Depictions: A Graph-based Understanding of Place Descriptions. Randell, D. A., Cui, Z. and Cohn, A. G. (1992). A Spatial Logic based on Regions and Connection. 3rd Int. Conf. on Knowledge Representation and Reasoning, Morgan Kaufmann. Vasardani, M., S. Timpf, S. Winter and M. Tomko (2013). From descriptions to depictions: A conceptual framework. Spatial information theory, Springer