Efficient zero shot event extraction with context definition

An apparatus comprising:

• at least one memory configured to store program code;
• and
• at least one processor configured to read the program code and operate as instructed by the program code, the program code comprising:
• training code configured to cause the at least one processor to train a context encoder, a first definition encoder, and a second definition encoder with auto extracted context-definition alignment data;
• retrieving code configured to cause the at least one processor to retrieve a plurality of verbal synsets from a lexical database;
• refining code configured to cause the at least one processor to refine a representation model based on the context-definition alignment data and the plurality of verbal synsets;
• first encoding code configured to cause the at least one processor to encode a plurality of candidate event type definitions;
• second encoding code configured to cause the at least one processor to encode the refined representation model with the trained context encoder;
• and
• determining code configured to cause the at least one processor to determine whether the encoded representation model belongs to one of the plurality of candidate event type definitions based on a cosine similarity between the encoded representation model, the trained context encoder, the first trained definition encoder, and the second trained definition encoder.

A non-transitory computer-readable storage medium, storing instructions, which, when executed by at least one processor, cause the at least one processor to:

• train a context encoder, a first definition encoder, and a second definition encoder with auto extracted context-definition alignment data;
• retrieve a plurality of verbal synsets from a lexical database;
• refine a representation model based on the context-definition alignment data and the plurality of verbal synsets;
• encode a plurality of candidate event type definitions;
• encode the refined representation model with the trained context encoder;
• and
• determine whether the encoded representation model belongs to one of the plurality of candidate event type definitions based on a cosine similarity between the encoded representation model, the trained context encoder, the first trained definition encoder, and the second trained definition encoder.