Automated contextual information retrieval based on multi-tiered user modeling and dynamic retrieval strategy

Patent metadata

Field	Value
Publication number	US-9633140-B2
Application number	US-201113024467-A
Country	US
Kind code	B2
Filing date	Feb 10, 2011
Priority date	Feb 10, 2011
Publication date	Apr 25, 2017
Grant date	Apr 25, 2017

Abstract

Automated contextual information retrieval techniques are provided based on multi-tiered user modeling and a dynamic retrieval strategy. Content relevant to a current message is presented by initially obtaining a multi-tiered user model containing a multi-tiered representation of interactions of a first user with each contact, wherein the multi-tiered representation includes a plurality of topic models each corresponding to interactions between the first user and one contact. The topic models contain a set of topics, each containing topic keywords. Context information is extracted based on content of the current message, a sender and/or a recipient of the current message, and the multi-tiered user model. A retrieval strategy is determined based on the extracted context information. Contextual queries are generated to search the information repositories selected based on the determined retrieval strategy. Content relevant to the current message is presented based on search results from the selected information repositories.

First claim

What is claimed is:

1. A method for automatically presenting content relevant to a current message, said method comprising:
obtaining a multi-tiered user model containing a multi-tiered representation of pair-wise interactions of a first user with each of one or more contacts, wherein the multi-tiered representation includes a plurality of topic models, wherein each of the topic models is generated from a pair-wise interaction between the first user and one of the contacts for at least one communication outside of a communication thread of said current message, wherein each of the topic models contains a set of topics, wherein each topic is associated with (i) a first probability of a word given the topic for substantially all words, wherein the first probability provides a list of topic keywords that describe the topic, and (ii) a second probability of the topic given a message for substantially all messages in the associated pair-wise interaction, wherein the second probability provides a list of messages associated with the topic;
extracting context information of a current message based on a content of the current message, one or more of a sender and a recipient of the current message, and the multi-tiered user model;
determining the topic of the current message by computing a topic distribution of the current message based on the second probabilities for each topic and selecting the topic with the highest probability;
determining a retrieval strategy based on the extracted context information of the current message;
selecting from a set of information repositories for contextual information retrieval based on the determined retrieval strategy, wherein the set of information repositories comprises one or more of people directories, a local memory, one or more online repositories, an email repository and a calendar entry repository, wherein at least one information repository in said set is outside of said communication thread of said current message;
generating one or more contextual queries for document retrieval based on one or more words contained in the current message and one or more keywords associated with the determined topic of the current message to search the selected information repositories; and
presenting the content relevant to the current message from the selected information repositories based on the one or more contextual queries.

2. The method of claim 1 , wherein the contacts comprise one or more of email contacts, calendar contacts, groups of email contacts and groups of calendar contacts.

3. The method of claim 1 , wherein the multi-tiered user model is obtained by extracting and aggregating information from email and calendar content of the first user, and applying statistical techniques to create topic models corresponding to pair-wise interactions of the first user with each of the contacts of said first user.

4. The method of claim 1 , wherein the current message comprises one or more of an email message, a text message and a transcribed voice mail message.

5. The method of claim 1 , wherein the contextual queries to search the selected information repositories are determined based on the content of the current message, the topic keywords contained in a determined topic of the current message, one or more of the sender and recipient of the current message, and the determined retrieval strategy.

6. The method of claim 1 , further comprising the step of selecting one or more topic models from the multi-tiered user model based on one or more of a sender and recipient of the current message, and matching the content of the current message to the topics of the selected topic models to find the best topic.

7. The method of claim 1 , wherein the content relevant to the current message is determined by processing the search results from the selected information repositories based on the context of the current message and the determined retrieval strategy.

8. A system for automatically presenting content relevant to a current message, said system comprising:
a memory; and
at least one processor, coupled to the memory, operative to:
obtain a multi-tiered user model containing a multi-tiered representation of pair-wise interactions of a first user with each of one or more contacts, wherein the multi-tiered representation includes a plurality of topic models, wherein each of the topic models is generated from a pair-wise interaction between the first user and one of the contacts for at least one communication outside of a communication thread of said current message, wherein each of the topic models contains a set of topics, wherein each topic is associated with (i) a first probability of a word given the topic for substantially all words, wherein the first probability provides a list of topic keywords that describe the topic and (ii) a second probability of the topic given a message for substantially all messages in the associated pair-wise interaction, wherein the second probability provides a list of messages associated with the topic;
extract context information of a current message based on a content of the current message, one or more of a sender and a recipient of the current message, and the multi-tiered user model;
determine the topic of the current message by computing a topic distribution of the current message based on the second probabilities for each topic and selecting the topic with the highest probability;
determine a retrieval strategy based on the extracted context information of the current message;
select from a set of information repositories for contextual information retrieval based on the determined retrieval strategy, wherein the set of information repositories comprises one or more of people directories, a local memory, one or more online repositories, an email repository and a calendar entry repository, wherein at least one information repository in said set is outside of said communication thread of said current message;
generate one or more contextual queries for document retrieval based on one or more words contained in the current message and one or more keywords associated with the determined topic of the current message to search the selected information repositories; and
present the content relevant to the current message from the selected information repositories based on the one or more contextual queries.

9. The system of claim 8 , wherein the multi-tiered user model is obtained by extracting and aggregating information from email and calendar content of the first user, and applying statistical techniques to create topic models corresponding to pair-wise interactions of the first user with each of the contacts of said first user.

10. The system of claim 8 , wherein the current message comprises one or more of an email message, a text message and a transcribed voice mail message.

11. The system of claim 8 , wherein the contextual queries to search the selected information repositories are determined based on the content of the current message, the topic keywords contained in a determined topic of the current message, one or more of the sender and recipient of the current message, and the determined retrieval strategy.

12. The system of claim 8 , wherein said processor is further configured to select one or more topic models from the multi-tiered user model based on one or more of a sender and recipient of the current message, and matching the content of the current message to the topics of the selected topic models to find the best topic.

13. The system of claim 8 , wherein the content relevant to the current message is determined by processing the search results from the selected information reposito

Assignees

• Lai Jennifer
• Lu Jie
• Pan Shimei
• Wen Zhen
• IBM

Inventors

• Lai Jennifer
• Lu Jie
• Pan Shimei
• Wen Zhen

Classifications

• H04L51/02

  using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages · CPC title

• G06F17/30997 Primary

  Physics · mapped topic

• G06F16/907 Primary

  Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually · CPC title

• G06F16/90335 Primary

  Query processing · CPC title