Nippon Telegraph and Telephone Corporation

Sony Group Corporation

Ntt Corporation

Masaaki Nagata

Tsutomu Hirao

Jun Suzuki

Masaaki Nishino

Michael Siegfried Spranger

Sho Takase

Makoto Morishita

Jin Nakayama

Kojiro Kashiwa

Hidetaka Kamigaito

Katsuyoshi Kanemoto

Akihito Kumakura

Tomohiro Nakatani

Yasuhiro Matsuda

Atsunori Ogawa

Yohei Yamanaka

Ryota Ando

Nobuo Sato

Koichiro Shimoda

Ippei Murofushi

Yoshinori Akisawa

Shinsaku Sakaue

An information learning apparatus includes a memory and a processor configured to perform generating, for each of processing units constituting an input sequence included in training data, a third embedded vector based on a first embedded vector for the processing unit and a second embedded vector corresponding to an unknown word; executing a process based on a learning target parameter, with the third embedded vector generated for each of the processing units as an input; and learning, for a processing result by the executing, the parameter based on an error of an output corresponding to the input sequence in the training data.

Embedding an unknown word in an input sequence

This disclosure relates to a device, a method, and a program capable of removing erroneous data from learning data used for machine learning used in natural language processing, for example. The method includes storing a forward direction learned model of a discrete series converter. The model is trained based on a plurality of pairs of discrete series of texts. Each pair comprises a first discrete series indicates an input of discrete series. A second discrete series indicates an output of discrete series. The first discrete series and the second discrete series are correctly associated. The method further includes converting the first discrete series to the second discrete series, and generating a quality score using the forward direction learned model, using a second learning pair of discrete series texts including an error in relationship.

Learning quality estimation device, method, and program

Included are input means for inputting first data that is data relating to a plurality of letters included in a text string that is a generation target, and generating means for generating second data that is data relating to the text string that satisfies predetermined constraint conditions including at least a condition relating to plausibility of the sequence of letters, on the basis of the first data.

Generation apparatus, generation method and program

There is provided an information processing apparatus which includes a creation unit and a processing unit. The creation unit is configured to create support information for supporting at least one of construction and maintenance of a database on the basis of first information including a representative question and an answer sentence associated with the representative question and stored in the database stored in a storage unit and second information indicating a history of reception with respect to a user. The processing unit is configured to execute processing of performing at least one of construction and maintenance of the database on the basis of input information input to the support information created by the creation unit.

Information processing apparatus and information processing method

Provided are an information processing apparatus and an information processing method capable of supporting work of giving a classification to a group of sentences. The information processing apparatus includes a presentation unit that presents a sentence included in a cluster of interest among clusters generated by clustering a sentence set in a sentence selection region, and a reception unit that receives selection of the sentence constituting a group of sentences from the sentences presented in the sentence selection region.

Classification of sentences from clusters of interest

To generate a vector representing a word in consideration of the importance of the character string of the word. A character N-gram vector acquisition unit acquires a vector representing each of character N-grams included in a word. A weight calculation unitcalculates a weight for a vector representing each of the character N-grams on the basis of the vector representing each of the character N-grams included in the word, and a character N-gram vector combining unitgenerates a composite vector obtained by combining the vector representing each of the character N-grams on the basis of the vector representing each of the character N-grams included in the word and the weight for the vector representing each of the character N-grams.

Word coding device, analysis device, language model learning device, method, and program

This disclosure relates to a method of generating a symbol string based on an input sentence represented by a sequence of symbols. In particular, the method involves receiving an input symbol string representing a sentence, generating, using a neural network based on a sequence of dependency structure of elements in the input symbol string, an output symbol string corresponding to the input sentence. The neural network includes an encoder that converts elements of the input symbol string to a first hidden state in a form of a multi-dimensional vector, an attention mechanism that applies a weight to the first hidden state and generates the weighted first hidden state as a second hidden state, a decoder that generates a third hidden state based on at least one element of the input symbol string, an element of the output symbol string, and the second hidden state, and an output generator that generates an element of the output symbol string based on the second hidden state and the third hidden state. The output symbol string may represent a sequence of labels to process the input sentence.

Symbol sequence generation apparatus, text compression apparatus, symbol sequence generation method and program

A speech recognition unit converts an input utterance sequence into a confusion network sequence constituted by a k-best of candidate words of speech recognition results; a lattice generating unit generates a lattice sequence having the candidate words as internal nodes and a combination of k words among the candidate words for an identical speech as an external node, in which edges are extended between internal nodes other than internal nodes included in an identical external node, from the confusion network sequence; an integer programming problem generating unit generates an integer programming problem for selecting a path that maximizes an objective function including at least a coverage score of an important word, of paths following the internal nodes with the edges extended, in the lattice sequence; and the summary generating unit generates a high-quality summary having less speech recognition errors and low redundancy using candidate words indicated by the internal nodes included in the path selected by solving the integer programming problem, under a constraint on the length of a summary to be generated.

Abstract generation device, method, program, and recording medium

The present disclosure relates to a method of evaluating accuracy of a summary of a document. The method includes receiving a plurality of reference summaries of a document and a system summary of the document. The system summary is generated by a machine. The method further includes extracting, for each reference summary, a tuple that is a pair of words composed of a modified word and a dependent word having a dependency relation to the modified word and a label representing the dependency relation. The method further includes replacing, for each of the extracted tuples, each of the modified word of the tuple's word pair and the dependent word with a class predetermined for the words. The method further generates a score of the system summary based on the class and a set of tuples of the system summary.

Summary evaluation device, method, program, and storage medium

The present invention relates to a method of accurately evaluating a system summary of a document based on one or more predefined units. The method includes dividing the document and the system summary of the document into the one or more predefined units, sentences and phrases, for example. For each of reference summaries for the document, the method generates an oracle. The oracle is a partial set of units that meet a length limitation and maximize a score generated by an evaluation function for a partial set of the units of the document based on the reference summary. The method further includes determines, based on the generated oracle, scores of the respective units included in the set of the oracle. The method further includes determining the score of the system summary based on the score of the system summary unit.

Summary evaluation device, method, program and storage medium

The present invention relates to summarization of a document. The method includes receiving texts of the document and a submodular function for defining an amount of information of a subset of the document as a summary of the document, generating a dependency structure tree representing a dependency structure of sentences in a document, identifying a set of paths including the root of the dependency structure tree, recursively selecting a paths using a greedy algorithm for maximizing the submodular function, and outputs a subset with the maximum amount of information as a the summary of the document.

Growing community of inventors

Tokyo, Japan

Masaaki Nagata