Microsoft Technology Licensing, LLC

Min Zeng

Ruihua Song

Yue Liu

Pingping Lin

Yan Chen

Li Zhou

Di Li

Lei Ding

Wei Liu

Xiang Xu

Shuang Peng

Bowen Wu

Baoxun Wang

Chao Zou

Ying Zou

Sa Kang

Feng Zhou

Methods and apparatuses for performing context completion to messages in a session are provided in the present disclosure. A message may be obtained. It may be detected that there exists context ellipsis in the message. It may be determined whether the message is retained in the current domain of the session. In response to determining that the message is retained in the current domain, a complementary text for recovering the context ellipsis may be selected in the current domain. A completed message may be generated based on the message and the complementary text.

Domain context ellipsis recovery for chatbot

Disclosed are a method and apparatus for automatically adding sound effects into an audio file. At least one trigger word may be identified from text content corresponding to the audio file (). Context information of the at least one trigger word may be determined from the text content (). It may be decided, based on the context information of the at least one trigger word, that sound effect addition is applicable the at least one trigger word (). A sound effect corresponding to the at least one trigger word may be selected from a sound effect library (). The sound effect may be integrated into the audio file ().

Automatically adding sound effects into audio files

The present disclosure provides method and apparatus for guiding topic in a conversation between a user and a chat engine. At least one first topic is determined. A first message is provided to the user based on the at least one first topic, to guide the conversation to the at least one first topic. A first response to the first message is received from the user. It is determined whether the first response is associated with the at least one first topic. In the case of determining that the first response is associated with the at least one first topic, at least one second topic is determined based on the at least one first topic. At least one second message is provided based at least on the at least one second topic, wherein if the at least one second topic is associated with resource or service, the at least one second message includes at least the resource or service.

Topic guiding in a conversation

The present disclosure provides a technical solution related to establishing a characterized chatbot with personality. On one hand, a corpus database matched with a specific character may be established, and a generated regular/conventional response messages may be converted into a characterized response messages with character's features during conversation, so that the conversation style of a chatbot may be characterized with a certain character's features. On the other hand, a chatbot may select a response message with a specific emotion by using an emotion conversation table corresponding to a personality when generating the response message against a content of a conversation input by a user, so that a chatbot may be characterized and have certain personality.

Characterized chatbot with personality

Non-limiting examples of the present disclosure describe a convolutional neural network (CNN) architecture configured to evaluate conversational relevance of query-response pairs. A CNN model is provided that can include a first branch, a second branch, and multilayer perceptron (MLP) layers. The first branch includes convolutional layers with dynamic pooling to process a query. The second branch includes convolutional layers with dynamic pooling to process candidate responses for the query. The query and the candidate responses are processed in parallel using the CNN model. The MLP layers are configured to rank query-response pairs based on conversational relevance.

Conversational relevance modeling using convolutional neural network

Growing community of inventors

Redmond, WA, United States of America

Min Zeng