Embedded Files
A central interest of my entire career has been concerned with increasing the scale of problems that can be solved by AI often by finding inspiration in database techniques needed to manage the ever increasing volume of data. A natural complement has been to look to AI methods for inspiration to solve problems in data management. A natural consequence has been work in data integration.
An unintended consequence is that, architecturally, I did a number of hybrid projects that today would be recognized as the neuro-symbolic design pattern. Since these projects predated the emergence of neural-network solutions other statistical methods were used and predate the term "neuro-symbolic".
Proto-Neuro-Symbolic Systems
Proto-Neuro-Symbolic Systems
Barbançon & Miranker (2007) — "SPHINX: Schema Integration by Example," Journal of Intelligent Information Systems, Vol. 29, pp. 145–184.
Version space symbolic learner guided by catalog-derived information gain statistics for active learning example selection.
Obermeyer & Miranker (1997) — "Evaluating Triggers Using Decision Trees," Proceedings of the Sixth ACM International Conference on Information and Knowledge Management (CIKM '97), pp. 144–150.
Database catalog selectivity statistics construct a decision tree that organizes symbolic ECA trigger condition evaluation.
Brant & Miranker (1993) — "Index Support for Rule Activation," ACM SIGMOD Record, pp. 42–48.
Statistical index structures guide symbolic production rule activation — the direct precursor pattern, with TREAT as the earlier motivating work.
Database research inspired at least in part by AI methods
AI research inspired at least in part by database methods
Ultrawrap: SPARQL execution on relational data, J. F. Sequeda and Daniel P. Miranker, Journal of Web Semantics, Vol. 22, 2013, pp. 19-39.
"An algorithmic basis for integrating production systems and large databases," D. P. Miranker and D. A. Brant, [1990] Proceedings. Sixth International Conference on Data Engineering, Los Angeles, CA, USA, 1990, pp. 353-360, doi: 10.1109/ICDE.1990.113488.
"The organization and performance of a TREAT-based production system compiler," D. P. Miranker and B. J. Lofaso, in IEEE Transactions on Knowledge and Data Engineering, vol. 3, no. 1, pp. 3-10, March 1991, doi: 10.1109/69.75882.
Data Integration
An unsupervised instance matcher for schema-free RDF data. J. Web Semant. 35: 102-123 (2015)
Schema matching over relations, attributes, and data values. SSDBM2014: 28:1-28:12
An Unsupervised Algorithm for Learning Blocking Schemes. ICDM 2013: 340-349
SPHINX: Schema integration by example. J. Intell. Inf. Syst. 29(2): 145-184(2007)
A Tractable Query Cache by Approximation. SARA 2002: 14
Page updated
Google Sites
Report abuse