Conditional Random Fields for XML Trees

Florent Jousse (1, 2, 3) , Rémi Gilleron (1, 2, 3) , Isabelle Tellier (1, 2, 3) , Marc Tommasi (1, 2, 3)

1 MOSTRARE - Modeling Tree Structures, Machine Learning, and Information Extraction
2 LIFL - Laboratoire d'Informatique Fondamentale de Lille
3 GRAppA - LIFL - Groupe de Recherche en Apprentissage Automatique

We present Conditional Random Fields (XCRFs), a framework for building conditional models to label XML data. XCRFs are Conditional Random Fields over unranked trees (where every node has an unbounded number of children). The maximal cliques of the graph are triangles consisting of a node and two adjacent children. We equip XCRFs with efficient dynamic programming algorithms for inference and parameter estimation. We experiment XCRFs on tree labeling tasks for structured information extraction and schema matching. Experimental results show that labeling with XCRFs is suitable for these problems.

Mots clés: Graphical Models; Machine Learning; XML; Information Extraction; Labeling; Schema Matching

Domaines: Apprentissage [cs.LG]

Florent Jousse, Rémi Gilleron, Isabelle Tellier, Marc Tommasi. Conditional Random Fields for XML Trees. Workshop on Mining and Learning in Graphs, Sep 2006, Berlin, Germany. ⟨inria-00118761⟩