公开(公告)号：WO2017201195A9

公开(公告)日：2017-11-23

申请号：PCT/US2017/033159

申请日：2017-05-17

Applicant: MALUUBA INC.

Inventor： TRISCHLER, Adam ,  BACHMAN, Philip ,  YUAN, Xingdi ,  YE, Zheng ,  SORDONI, Alessandro

IPC: G06F17/27

Abstract: Described herein are systems and methods for providing a natural language comprehension system that employs a two-stage process for machine comprehension of text. The first stage indicates words in one or more text passages that potentially answer a question. The first stage outputs a set of candidate answers for the question, along with a first probability of correctness for each candidate answer. The second stage forms one or more hypotheses by inserting each candidate answer into the question and determines whether a sematic relationship exists between each hypothesis and each sentence in the text. The second processing circuitry generates a second probability of correctness for each candidate answer and combines the first probability with the second probability to produce a score that is used to rank the candidate answers. The candidate answer with the highest score is selected as a predicted answer.

公开(公告)号：WO2017161189A1

公开(公告)日：2017-09-21

申请号：PCT/US2017/022812

申请日：2017-03-16

Applicant: MALUUBA INC. ,  BACHMAN, Philip

Inventor： BACHMAN, Philip ,  TRISCHLER, Adam ,  YE, Zheng ,  YUAN, Xingdi

IPC: G06F17/28 ,  G06N3/02 ,  G06N99/00

CPC classification number: G06N3/08 ,  G06F17/277 ,  G06F17/2785 ,  G06N3/0454 ,  G06N5/022 ,  G06N5/04 ,  G06N20/10

Abstract: Examples of the present disclosure provide systems and methods relating to a machine comprehension test with a learning-based approach, harnessing neural networks arranged in a parallel hierarchy. This parallel hierarchy enables the model to compare the passage, question, and answer from a variety of perspectives, as opposed to using a manually designed set of features. Perspectives may range from the word level to sentence fragments to sequences of sentences, and networks operate on word-embedding representations of text. A training methodology for small data is also provided.

Abstract translation: 本公开的示例提供了与利用基于学习的方法的机器理解测试相关的系统和方法，利用了排列在并行分层结构中的神经网络。 这种平行的层次结构使模型能够从各种角度比较段落，问题和答案，而不是使用手动设计的一组功能。 透视可以从单词级别到句子片段到句子序列，并且网络对文本的词语嵌入表示进行操作。 还提供了小数据的培训方法。

公开(公告)号：WO2017210634A1

公开(公告)日：2017-12-07

申请号：PCT/US2017/035812

申请日：2017-06-02

Applicant: MALUUBA INC.

Inventor： SORDONI, Alessandro ,  BACHMAN, Philip ,  TRISCHLER, Adam

IPC: G06F17/28 ,  G06N3/04

Abstract: Described herein are systems and methods for providing a natural language comprehension system (NLCS) that iteratively performs an alternating search to gather information that may be used to predict the answer to the question. The NLCS first attends to a query glimpse of the question, and then finds one or more corresponding matches by attending to a text glimpse of the text.

Abstract translation: 这里描述的是用于提供自然语言理解系统（NLCS）的系统和方法，该系统和方法迭代地执行交替搜索以收集可用于预测问题答案的信息。 NLCS首先查询问题的一瞥，然后通过关注文本的文本来查找一个或多个相应的匹配。

公开(公告)号：WO2017210613A1

公开(公告)日：2017-12-07

申请号：PCT/US2017/035767

申请日：2017-06-02

Applicant: MALUUBA INC.

Inventor： SHARMA, Shikhar ,  HE, Jing ,  SULEMAN, Kaheer ,  BACHMAN, Philip ,  SCHULZ, Hannes

IPC: G06F17/28

Abstract: Described herein are systems and methods for providing a natural language generator in a spoken dialogue system that considers both lexicalized and delexicalized dialogue act slot-value pairs when translating one or more dialogue act slot-value pairs into a natural language output. Each slot and value associated with the slot in a dialogue act are represented as (dialogue act + slot, value), where the first term (dialogue act + slot) is delexicalized and the second term (value) is lexicalized. Each dialogue act slot-value representation is processed to produce to produce at least one delexicalized sentence as an output. A lexicalized sentence is produced by replacing each delexicalized slot with the value associated with the delexicalized slot.

Abstract translation: 这里描述的是用于在口述对话系统中提供自然语言生成器的系统和方法，该系统和方法在将一个或多个对话动作时隙 - 值对转换为一个或多个对话动作时隙值对时考虑了词汇化和非动词化对话两者 自然语言输出。 与对话动作中的时隙相关联的每个时隙和值被表示为（对话动作+时隙，值），其中第一项（对话动作+时隙）被去灵活化并且第二项（值）被词汇化。 每个对话行为时隙值表示被处理以产生至少一个作为输出的delexicalized句子。 一个词汇化的句子是通过用与delexicalized插槽关联的值替换每个delexicalized插槽来生成的。

公开(公告)号：WO2017201195A1

公开(公告)日：2017-11-23

申请号：PCT/US2017/033159

申请日：2017-05-17

Applicant: MALUUBA INC.

Inventor： TRISCHLER, Adam ,  BACHMAN, Philip ,  YUAN, Xingdi ,  YE, Zheng ,  SORDONI, Alessandro

IPC: G06F17/27

Abstract: Described herein are systems and methods for providing a natural language comprehension system that employs a two-stage process for machine comprehension of text. The first stage indicates words in one or more text passages that potentially answer a question. The first stage outputs a set of candidate answers for the question, along with a first probability of correctness for each candidate answer. The second stage forms one or more hypotheses by inserting each candidate answer into the question and determines whether a sematic relationship exists between each hypothesis and each sentence in the text. The second processing circuitry generates a second probability of correctness for each candidate answer and combines the first probability with the second probability to produce a score that is used to rank the candidate answers. The candidate answer with the highest score is selected as a predicted answer.

Abstract translation: 这里描述的是用于提供自然语言理解系统的系统和方法，该自然语言理解系统采用用于机器理解文本的两阶段过程。 第一阶段指出可能回答问题的一个或多个文本段落中的单词。 第一阶段输出一组针对该问题的候选答案，以及每个候选答案的第一个正确概率。 第二阶段通过将每个候选答案插入问题形成一个或多个假设，并确定每个假设与文本中每个句子之间是否存在语义关系。 第二处理电路针对每个候选答案生成第二正确率概率，并将第一概率与第二概率组合以产生用于对候选答案进行排序的分数。 具有最高分数的候选答案被选为预测答案。