A5018546368- Natural Language Processing Techniques
- Topic Modeling
- Text and Document Classification Technologies
- AI-based Problem Solving and Planning
- Machine Learning and Algorithms
- Data Mining Algorithms and Applications
- Web Data Mining and Analysis
- Semantic Web and Ontologies
- Text Readability and Simplification
- Data Quality and Management
- Speech and dialogue systems
Google (United States)
2009-2015
We introduce a novel precedence reordering approach based on dependency parser to statistical machine translation systems.Similar other preprocessing approaches, our method can efficiently incorporate linguistic knowledge into SMT systems without increasing the complexity of decoding.For set five subject-object-verb (SOV) order languages, we show significant improvements in BLEU scores when translating from English, compared state-of-the-art phrase-based systems.
We present Plato, a probabilistic model for entity resolution that includes novel approach handling noisy or uninformative features, and supplements labeled training data derived from Wikipedia with very large unlabeled text corpus. Training inference in the proposed can easily be distributed across many servers, allowing it to scale over 10 7 entities. evaluate Plato on three standard datasets resolution. Our achieves best results to-date TAC KBP 2011 is highly competitive both CoNLL 2003...
We describe SLING, a framework for parsing natural language into semantic frames. SLING supports general transition-based, neural-network with bidirectional LSTM input encoding and Transition Based Recurrent Unit (TBRU) output decoding. The model is trained end-to-end using only the text tokens as input. transition system has been designed to frame graphs directly without any intervening symbolic representation. includes an efficient scalable store implementation well neural network JIT...