US 11,681,870 B2
Reducing latency and improving accuracy of work estimates utilizing natural language processing
Jeremy Bowers, Downers Grove, IL (US); and David Pearson, Downers Grove, IL (US)
Assigned to ENSONO, LP, Downers Grove, IL (US)
Appl. No. 16/963,882
Filed by Ensono, LP, Downers Grove, IL (US)
PCT Filed Jan. 28, 2019, PCT No. PCT/US2019/015447
§ 371(c)(1), (2) Date Jul. 22, 2020,
PCT Pub. No. WO2019/148115, PCT Pub. Date Aug. 1, 2019.
Claims priority of provisional application 62/622,559, filed on Jan. 26, 2018.
Prior Publication US 2021/0050019 A1, Feb. 18, 2021
Int. Cl. G06F 40/20 (2020.01); G10L 15/22 (2006.01); G10L 15/34 (2013.01)
CPC G06F 40/20 (2020.01) [G10L 15/22 (2013.01); G10L 15/34 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A system for reducing latency and improving accuracy in estimating work utilizing natural language processing devices and serverless event-driven parallel data processing architectures, the system comprising:
a project management platform configured to store information about one or more completed work projects in a database of completed stories and story point values, wherein the completed stories comprise keywords or phrases;
a natural language input and output device configured to capture audio and video input from a user and output an audio or video message including a work estimate, wherein the natural language input and output device comprises a speech engine that controls capture of an audio and video data stream upon detection of a wake word, and a communication interface configured to establish a communication channel between the natural language input and output device and a voice processing service, wherein the natural language input and output device transmits the audio and video data stream to the voice processing service over the communication channel, and wherein the voice processing service is configured to process the audio and video data stream to identify a command and one or more parameters, and create one or more response messages for transmitting to the natural language input and output device, and when the command corresponds to a request for the work estimate the voice processing service transmits an event trigger and the one or more parameters; and
a serverless compute service configured to receive application code, and the one or more parameters transmitted from the voice processing service, wherein the application code executes in response to the event trigger, the application code comprising instructions to determine the work estimate based on the one or more parameters,
wherein:
the one or more parameters define a scope of the work estimate;
the work estimate is determined by applying a Bayesian classifier to the one or more parameters and the information about one or more completed work projects stored in the project management platform;
the work estimate is further determined by calculating a prior probability for each of the keywords or phrases occurring in the completed stories;
the prior probabilities for the keywords or phrases are stored in a historical database;
applying the Bayesian classifier to the one or more parameters comprises classifying the one or more parameters based on the prior probabilities for the keywords or phrases;
the keywords or phrases occurring in the completed stories are dynamically displayed in a visualization module as a word cloud;
the word cloud displays positions of the keywords or phrases based on a clustering the keywords or phrases by relevance distances representing semantic closeness; and
the relevance distances are generated based on multidimensional scaling of a matrix in an n-dimensional space that represents distances between the keywords or phrases.