US 11,942,086 B2
Description support device and description support method
Natsuki Saeki, Kanagawa (JP); Shoichi Araki, Osaka (JP); Masakatsu Hoshimi, Osaka (JP); and Takahiro Kamai, Kyoto (JP)
Assigned to PANASONIC INTELLECTUAL PROPERTY MANAGEMENT CO., LTD., Osaka (JP)
Filed by Panasonic Intellectual Property Management Co., Ltd., Osaka (JP)
Filed on Dec. 17, 2020, as Appl. No. 17/125,295.
Application 17/125,295 is a continuation of application No. PCT/JP2019/036504, filed on Sep. 18, 2019.
Claims priority of application No. 2018-182534 (JP), filed on Sep. 27, 2018.
Prior Publication US 2021/0104240 A1, Apr. 8, 2021
Int. Cl. G10L 15/22 (2006.01); G06Q 30/016 (2023.01); G10L 15/06 (2013.01)
CPC G10L 15/22 (2013.01) [G10L 15/06 (2013.01); G06Q 30/016 (2013.01)] 17 Claims
OG exemplary drawing
 
1. A description support device for displaying information on a topic to be checked in an utterance by a user, the description support device comprising:
an inputter to sequentially acquire input information from the user, wherein the input information includes a plurality of utterance sentences;
a controller to receive a check result of the topic for each utterance sentence in the input information from a language processor;
a display to display the check result received by the controller;
a communication interface to communicate data with the language processor provided externally from the description support device; and
a memory to store a checklist indicating the topic to be checked,
wherein the controller is configured to cause the display to display the check result, based on the checklist stored in the memory, the check result indicating whether or not the topic in the checklist is described in each utterance sentence and a corresponding likelihood of the utterance sentence,
wherein the controller is configured to cause the communication interface to: automatically transmit the input information to the language processor in response to acquiring the input information from the inputter; and automatically receive the check result from the language processor in response to calculating the likelihood by the language processor, wherein the likelihood of each utterance sentence includes the likelihood indicating a degree to which each utterance sentence is predicted to be a keyword together with the likelihood of the topic in the checklist is detected in the corresponding utterance sentence, wherein the language processor is configured to calculate the likelihood of each utterance sentence by executing word embedding to convert each utterance sentence into a vector, input the vector into a learned model to search keyword and to check whether the topic in the checklist is detected in each utterance sentence and output the likelihood of each utterance sentence,
wherein the controller is configured to cause the display to display, according to the likelihood of each utterance sentence, the check result as display information.