Vocal utterance based item inventory actions

Patent metadata

Field	Value
Publication number	US-11004449-B2
Application number	US-201816204880-A
Country	US
Kind code	B2
Filing date	Nov 29, 2018
Priority date	Nov 29, 2018
Publication date	May 11, 2021
Grant date	May 11, 2021

Abstract

Methods, computer program products, and systems are presented. The method computer program products, and systems can include, for instance: obtaining vocal utterance data representing vocal utterances of multiple users within a venue; processing the vocal utterance data to return metadata associated to the vocal utterance data; predicting using the metadata an item for acquisition by one or more user of the multiple users; and returning an action decision in dependence on the predicting.

First claim

The invention claimed is:

1. A computer implemented method comprising:
obtaining, using one or more audio input device, vocal utterance data representing vocal utterances of multiple users within a venue;
processing the vocal utterance data representing vocal utterances of the multiple users within the venue to return metadata associated to the vocal utterance data representing vocal utterances of the multiple users within the venue;
predicting using the metadata items for acquisition by respective ones of the multiple users; and
returning action decisions in dependence on the predicting.

2. The computer implemented method of claim 1 , wherein the action decision includes an action to activate an interactive session between a voice enabled device and a user of the multiple users, wherein vocal utterance responses of the user are captured by the voice enabled device and processed in response to synthesized voice prompts presented by the voice enabled device.

3. The computer implemented method of claim 1 , wherein the action decision includes an action to activate automated delivery of the item to the venue by an autonomous vehicle.

4. The computer implemented method of claim 1 , wherein the action decision includes an action to activate an automated social media post on behalf of a user of the multiple users, the social media post referencing the item for acquisition.

5. The computer implemented method of claim 1 , wherein the predicting includes predicting using the metadata a plurality of different items for acquisition by one or more user of the multiple users, and wherein the returning an action decision includes returning differentiated action decisions in dependence on the predicting, the differentiated action decisions being differentiated in dependence on: (a) respective fair market values associated to the different items for acquisition; (b) respective confidence levels associated to respective predictions of item acquisitions for the respective different items, and (c) respective probabilities of acquisition associated to respective predictions of item acquisitions for the respective different items, wherein the differentiated action decisions include an action to activate an interactive session between a voice enabled device and a user of the multiple users, wherein vocal utterance responses of the user are captured by the voice enabled device and processed in response to synthesized voice prompts presented by the voice enabled device, an action to automatically initiate a purchase transaction of the item for acquisition, an action to activate automated delivery of the item to the venue by an autonomous vehicle, and an action to activate an automated social media post on behalf of a user of the multiple users, the social media post referencing the item for acquisition.

6. The computer implemented method of claim 1 , wherein the predicting includes using a predictive model, wherein the predictive model has been trained using iteratively applied sets of training data wherein respective sets of the iteratively applied sets of training data include: (a) data specifying a list of items acquired during a time period T=t; and (b) vocal utterance processing derived metadata for the time period T=t−1 preceding time period T=t.

7. The computer implemented method of claim 1 , wherein the predicting includes using a predictive model, wherein the predictive model has been trained using historical metadata associated to historical vocal utterances, and data of historical inventory item changes associated to the venue.

8. The computer implemented method of claim 1 , wherein the predicting includes using a first predictive model, and a second predictive model, wherein the first predictive model has been trained using iteratively applied sets of training data that include (a) data of observed item acquisitions for a venue for a certain time period combined with (b) utterance data derived metadata for a time period preceding the certain time period, wherein the second predictive model uses IOT sensor data to monitor a physical presence of items in an item storage location within the venue, and wherein the second predictive model has been configured to predict that an item inventory will be maintained at a constant baseline level.

9. The computer implemented method of claim 1 , wherein the method includes obtaining inventory change indicating data from multiple data sources, processing the inventory change indicating data to determine an inventory change in an item inventory associated to the venue, and applying the inventory change as training data for training a predictive model used for performance of the predicting, wherein the multiple data sources include (a) an enterprise system provided by an online retail store used by a user of the multiple users, and (b) an IOT sensor system that includes an IOT sensor device disposed to sense a physical presence of items within a storage location of the venue.

10. The computer implemented method of claim 1 , wherein the method includes obtaining inventory change indicating data from multiple data sources, processing the inventory change indicating data to determine an inventory change in an item inventory associated to the venue, and applying the inventory change as training data for training a predictive model used for performance of the predicting, wherein the multiple data sources include (a) client computer devices associated with respective ones of the multiple users within the venue, and an IOT sensor system having multiple IOT sensor devices distributed within the venue.

11. The computer implemented method of claim 1 , wherein the processing the vocal utterance data to return metadata associated to the vocal utterance data includes using processing to return non-speech event vocal utterance metadata associated to the vocal utterance data, the non-speech event vocal utterance metadata indicating an occurrence of a non-speech vocal utterance event.

12. The computer implemented method of claim 1 , wherein the processing the vocal utterance data to return metadata associated to the vocal utterance data includes using Natural Language Processing to return topic metadata associated to the vocal utterance data, wherein the processing the vocal utterance data to return metadata associated to the vocal utterance data includes using processing to return non-speech event vocal utterance metadata associated to the vocal utterance data, the non-speech event vocal utterance metadata indicating an occurrence of a non-speech vocal utterance event, wherein the predicting using the metadata includes predicting that a first item will be acquired for the venue in dependence on topic metadata of the metadata, and wherein the predicting using the metadata includes predicting that a second item will be acquired for the venue in dependence on non-speech event vocal utterance metadata of the metadata.

13. The computer implemented method of claim 1 , wherein the predicting includes predicting using the metadata a, plurality of different items for acquisition by one or more user of the multiple users, and wherein the returning an action decision includes returning differentiated action decisions in dependence on the predicting, the differentiated action decisions being differentiated in dependence on: (a) respective fair market values associated to the different items for acquisition; and (b) respective probabilities of acquisition associated to respective predictions of item acquisitions for the respective different items, wherein the differentiated action decisions include an action to activate an interactive session between a voice enabled device and a user

Assignees

• IBM

Inventors

• Bender Michael
• Fox Jeremy R
• Bhogal Kulvir

Classifications

• G06Q10/40

  Business processes related to social networking or social networking services · CPC title

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G10L2015/228

  of application context · CPC title

• G06N5/022

  Knowledge engineering; Knowledge acquisition · CPC title