International Business Machines Corporation

Yi Peng Yu

Jian Wang

Qi Cheng Li

Li Jun Mei

Xin Zhou

Hao Chen

Shao Chun Li

Ya Bin Dang

Jie Ma

David Nahamoo

Zhuoxuan Jiang

A computer-implemented method is presented for enabling hierarchical conversational policy learning for sales strategy planning. The method includes enabling a user to have a conversation with a robot via a conversation platform, employing a plan database to store general plans used in the conversation, employing an industry database to store a plurality of candidate plans pertaining to sales promotions, and employing a plan and policy optimizer to allow the robot to select and output an optimal plan from the plurality of candidate plans, the optimal plan determined by hierarchical reinforcement learning via a first learner and a second learner, the first leaner selecting the optimal plan and the second learner selecting an optimal action.

Hierarchical conversational policy learning for sales strategy planning

Computer-implemented methods, computer program products, and system facilitating evaluation of a progress of a diagnosis process are provided. The computer-implemented method comprises: obtaining, by a device operatively coupled to one or more processing units, a list of symptom(s); retrieving, by the device, a sub-graph associated with the list of symptom(s) from a knowledge graph; extracting, by the device, a list of feature(s) from the retrieved sub-graph; estimating, by the device, an average number of inquiries based on the extracted list of feature(s); and calculating, by the device, the progress of the diagnosis process based on the estimated average number of inquiries.

Progress evaluation of a diagnosis process

This disclosure provides a method for road condition prediction. The method comprises receiving, for at least one source vehicle, sensor data collected by a sensor associated with the source vehicle. The method further comprises identifying, based on the sensor data, at least a location with an abnormal road condition and a responsive action taken by the source vehicle. The method further comprises notifying, the location with the abnormal road condition and the responsive action to at least one target vehicle that is expected to pass the location. This disclosure also provides a computer system and a computer software product for road condition prediction.

Road condition prediction

A computer-implemented method, computer program product, and computer processing system are provided for generating inferences from a forest of predefined problem determination trees using a processor-based conversation platform. The method includes selecting a tree from among the forest of predefined problem determination trees, responsive to user utterances uttered during an inference generating session. The method further includes navigating the tree to allocate a relevant tree node to generate a problem diagnosis question or a problem resolution action by understanding the user utterances among common interaction patterns in problem diagnosis and problem resolution dialogs. The method also includes providing speech for uttering the problem diagnosis question or the problem resolution action to a user.

Forest inference engine on conversation platform

A cognitive conversation system that generates effective diagnostic questions is provided. The cognitive conversation system receives a set of currently known symptoms (or currently available answers to diagnostic questions) of a reported problem or fault. The system identifies (i) a set of possible root causes of the reported problem based on the currently known symptoms and (ii) probabilities for the set of possible root causes by using a bipartite graph data structure that links possible symptoms with possible root causes. Upon determining that at least one possible root cause has a probability that is higher than a threshold, the system presents an explanation or solution associated with the at least one possible root cause. Upon determining that none of the possible root causes in the set of possible root causes has a probability higher than the threshold, the system presents a question based on information entropy that is computed based on probabilities of the identified possible root causes.

Conversational problem determination based on bipartite graph

An embodiment of the invention may include a method, computer program product and computer system for human-machine communication. The method, computer program product and computer system may include a computing device that maps linguistic data of source content to a vector. The computing device may cluster the linguistic data of source content. The computing device may determine a plurality of segments based on the mapped linguistic data and the clustered linguistic data. The computing device may transform a segment of the plurality of segments into representative data, the representative data is a function of the remaining plurality of segments.

Mining procedure dialogs from source content

A computer-implemented method, apparatus, computer program product for intervention in a conversation between a virtual agent and a user is disclosed. In the computer-implemented method, a risk level of the conversation is evaluated. In response to the risk level being higher than a threshold, it is indicated that the conversation needs an intervention by a human agent. In response to the human agent intervening in the conversation, the conversation is handed over from the virtual agent to the human agent. Then, a simulative conversation is generated according to the conversation between the user and the human agent. In response to a determination by the human agent based on the simulative conversation, the conversation is handed over from the human agent to the virtual agent.

Growing community of inventors

Beijing, China

Yi Peng Yu