Semantic Parsing of Tamil Sentences
Balaji Jagan, T. V. Geetha, Ranjani Parthasarathi · 2012
In this paper, we propose a rule-based approach for the identificat ion of semantic sub-graphs from Tamil sentences. In order to achieve the goal of semantic sub -graph identification and construction, we use a semantic graph based representation called Univ ersal Networking Language (UNL), which is a directed acyclic graph representation. To identify and build the semantic sub-graphs, we classify the ru les based on morpho-semantic features which include word associated features and context based features . The rules are performed in two stages; one while build ing the simple UNL graphs and one after the simple UNL graphs construction. We have identified 18 rules for sub-graph identification and construction.