Abstract: A decoder includes a feature extraction circuit for calculating one or more feature vectors. An acoustic model circuit is coupled to receive one or more feature vectors from and assign one or more likelihood values to the one or more feature vectors. A memory architecture that utilizes on-chip state lattices and an off-chip memory for storing states of transition of the decoder is used to reduce reading and writing to the off-chip memory. The on-chip state lattice is populated with at least one of the states of transition stored in the off-chip memory. An on-chip word is generated from a snapshot from the on-chip state lattice. The on-chip state lattice and the on-chip word lattice act as an on-chip cache to reduce reading and writing to the off-chip memory.

Abstract: Methods and systems for natural language processing include pretraining a machine learning model that is based on a bidirectional encoder representations from transformers model, using a span selection training data set that associates a masked word with a passage. A natural language processing task is performed using the span selection pretrained machine learning model.

Abstract: A computer-implemented method for data labeling is provided. The computer-implemented method assigns pseudo-labels to unlabeled examples of data using a similarity metric on an embedding space to produce pseudo-labeled examples. A curriculum learning model is trained using the pseudo-labeled examples. The curriculum learning model trained with the pseudo-labeled examples is employed in in a fine-tuning task to enhance classification accuracy of the data.

Abstract: Methods and systems for natural language processing include pretraining a machine learning model that is based on a bidirectional encoder representations from transformers model, using a span selection training data set that associates a masked word with a passage. A natural language processing task is performed using the span selection pretrained machine learning model.

Abstract: A computer-implemented method for data labeling is provided. The computer-implemented method assigns pseudo-labels to unlabeled examples of data using a similarity metric on an embedding space to produce pseudo-labeled examples. A curriculum learning model is trained using the pseudo-labeled examples. The curriculum learning model trained with the pseudo-labeled examples is employed in in a fine-tuning task to enhance classification accuracy of the data.

Abstract: According to an aspect, learning parameters in a feed forward probabilistic graphical model includes creating an inference model via a computer processor. The creation of the inference model includes receiving a training set that includes multiple scenarios, each scenario comprised of one or more natural language statements, and each scenario corresponding to a plurality of candidate answers. The creation also includes constructing evidence graphs for each of the multiple scenarios based on the training set, and calculating weights for common features across the evidence graphs that will maximize a probability of the inference model locating correct answers from corresponding candidate answers across all of the multiple scenarios. In response to an inquiry from a user that includes a scenario, the inference model constructs an evidence graph and recursively constructs formulas to express a confidence of each node in the evidence graph in terms of its parents in the evidence graph.

Abstract: According to an aspect, learning parameters in a feed forward probabilistic graphical model includes creating an inference model via a computer processor. The creation of the inference model includes receiving a training set that includes multiple scenarios, each scenario comprised of one or more natural language statements, and each scenario corresponding to a plurality of candidate answers. The creation also includes constructing evidence graphs for each of the multiple scenarios based on the training set, and calculating weights for common features across the evidence graphs that will maximize a probability of the inference model locating correct answers from corresponding candidate answers across all of the multiple scenarios. In response to an inquiry from a user that includes a scenario, the inference model constructs an evidence graph and recursively constructs formulas to express a confidence of each node in the evidence graph in terms of its parents in the evidence graph.

Abstract: A method, system and computer program product for recognizing terms in a specified corpus. In one embodiment, the method comprises providing a set of known terms t?T, each of the known terms t belonging to a set of types ? (t)={?1, . . . }, wherein each of the terms is comprised of a list of words, t=w1, w2, . . . , wn, and the union of all the words for all the terms is a word set W. The method further comprises using the set of terms T and the set of types to determine a set of pattern-to-type mappings p??; and using the set of pattern-to-type mappings to recognize terms in the specified corpus and, for each of the recognized terms in the specified corpus, to recognize one or more of the types ? for said each recognized term.

Abstract: Embodiments provide a system and method for short form and long form detection. Using a language-independent process, the detection system can ingest a corpus of documents, pre-process those documents by tokenizing the documents and performing a part-of-speech analysis, and can filter one or more candidate short forms using one or more filters that select for semantic criteria. Semantic criteria can include the part of speech of a token, whether the token contains more than a pre-determined amount of symbols or digits, whether the token appears too frequently in the corpus of documents, and whether the token has at least one uppercase letter. The detection system can detect short forms independent of case and punctuation, and independent of language-specific metaphone variants.

Abstract: Embodiments provide a system and method for short form and long form detection. Using a language-independent process, the detection system can ingest a corpus of documents, pre-process those documents by tokenizing the documents and performing a part-of-speech analysis, and can filter one or more candidate short forms using one or more filters that select for semantic criteria. Semantic criteria can include the part of speech of a token, whether the token contains more than a pre-determined amount of symbols or digits, whether the token appears too frequently in the corpus of documents, and whether the token has at least one uppercase letter. The detection system can detect short forms independent of case and punctuation, and independent of language-specific metaphone variants.

Abstract: Embodiments provide a system and method for short form and long form detection. Given candidate short forms, the system can generate one or more n-gram combinations, resulting in one or more candidate short form and n-gram combination pairs. For each candidate short form and n-gram combination pair, the system can calculate an approximate string matching distance, calculate a best possible alignment score, calculate a confidence score, calculate a topic similarity score, and calculate a semantic similarity score. The system can determine the validity, through a meta learner, of the one or more valid candidate short form and n-gram combination pairs based upon each short form and n-gram combination pair's confidence score, topic similarity score, and semantic similarity score, and store the valid short form and n-gram combination pairs in a repository. The system has no language specific constraints and can extract short form and long form pairs from documents written in various languages.

Abstract: Embodiments provide a system and method for short form and long form detection. Using a language-independent process, the detection system can ingest a corpus of documents, pre-process those documents by tokenizing the documents and performing a part-of-speech analysis, and can filter one or more candidate short forms using one or more filters that select for semantic criteria. Semantic criteria can include the part of speech of a token, whether the token contains more than a pre-determined amount of symbols or digits, whether the token appears too frequently in the corpus of documents, and whether the token has at least one uppercase letter. The detection system can detect short forms independent of case and punctuation, and independent of language-specific metaphone variants.

Abstract: Embodiments provide a system and method for short form and long form detection. Given candidate short forms, the system can generate one or more n-gram combinations, resulting in one or more candidate short form and n-gram combination pairs. For each candidate short form and n-gram combination pair, the system can calculate an approximate string matching distance, calculate a best possible alignment score, calculate a confidence score, calculate a topic similarity score, and calculate a semantic similarity score. The system can determine the validity, through a meta learner, of the one or more valid candidate short form and n-gram combination pairs based upon each short form and n-gram combination pair's confidence score, topic similarity score, and semantic similarity score, and store the valid short form and n-gram combination pairs in a repository. The system has no language specific constraints and can extract short form and long form pairs from documents written in various languages.

Abstract: According to an aspect, a processing system of a question answering computer system determines a first set of relations between one or more pairs of terms in a question. The processing system also determines a second set of relations between one or more pairs of terms in a candidate passage including a candidate answer to the question. The processing system matches the first set of relations to the second set of relations. A plurality of scores is determined by the processing system based on the matching. The processing system aggregates the scores to produce an answer score indicative of a level of support that the candidate answer correctly answers the question.

Abstract: According to an aspect, a processing system of a question answering computer system determines a first set of relations between one or more pairs of terms in a question. The processing system also determines a second set of relations between one or more pairs of terms in a candidate passage including a candidate answer to the question. The processing system matches the first set of relations to the second set of relations. A plurality of scores is determined by the processing system based on the matching. The processing system aggregates the scores to produce an answer score indicative of a level of support that the candidate answer correctly answers the question.

Abstract: Embodiments provide a system and method for short form and long form detection. Using a language-independent process, the detection system can ingest a corpus of documents, pre-process those documents by tokenizing the documents and performing a part-of-speech analysis, and can filter one or more candidate short forms using one or more filters that select for semantic criteria. Semantic criteria can include the part of speech of a token, whether the token contains more than a pre-determined amount of symbols or digits, whether the token appears too frequently in the corpus of documents, and whether the token has at least one uppercase letter. The detection system can detect short forms independent of case and punctuation, and independent of language-specific metaphone variants.

Abstract: Embodiments provide a system and method for short form and long form detection. Given candidate short forms, the system can generate one or more n-gram combinations, resulting in one or more candidate short form and n-gram combination pairs. For each candidate short form and n-gram combination pair, the system can calculate an approximate string matching distance, calculate a best possible alignment score, calculate a confidence score, calculate a topic similarity score, and calculate a semantic similarity score. The system can determine the validity, through a meta learner, of the one or more valid candidate short form and n-gram combination pairs based upon each short form and n-gram combination pair's confidence score, topic similarity score, and semantic similarity score, and store the valid short form and n-gram combination pairs in a repository. The system has no language specific constraints and can extract short form and long form pairs from documents written in various languages.

Abstract: According to an aspect, learning parameters in a feed forward probabilistic graphical model includes creating an inference model via a computer processor. The creation of the inference model includes receiving a training set that includes multiple scenarios, each scenario comprised of one or more natural language statements, and each scenario corresponding to a plurality of candidate answers. The creation also includes constructing evidence graphs for each of the multiple scenarios based on the training set, and calculating weights for common features across the evidence graphs that will maximize a probability of the inference model locating correct answers from corresponding candidate answers across all of the multiple scenarios. In response to an inquiry from a user that includes a scenario, the inference model constructs an evidence graph and recursively constructs formulas to express a confidence of each node in the evidence graph in terms of its parents in the evidence graph.

Abstract: According to an aspect, learning parameters in a feed forward probabilistic graphical model includes creating an inference model via a computer processor. The creation of the inference model includes receiving a training set that includes multiple scenarios, each scenario comprised of one or more natural language statements, and each scenario corresponding to a plurality of candidate answers. The creation also includes constructing evidence graphs for each of the multiple scenarios based on the training set, and calculating weights for common features across the evidence graphs that will maximize a probability of the inference model locating correct answers from corresponding candidate answers across all of the multiple scenarios. In response to an inquiry from a user that includes a scenario, the inference model constructs an evidence graph and recursively constructs formulas to express a confidence of each node in the evidence graph in terms of its parents in the evidence graph.

Abstract: According to an aspect, a processing system of a question answering computer system determines a first set of relations between one or more pairs of terms in a question. The processing system also determines a second set of relations between one or more pairs of terms in a candidate passage including a candidate answer to the question. The processing system matches the first set of relations to the second set of relations. A plurality of scores is determined by the processing system based on the matching. The processing system aggregates the scores to produce an answer score indicative of a level of support that the candidate answer correctly answers the question.