Browse Prior Art Database

Recommendation of auto-reply messages based on user's reply pattern analysis Disclosure Number: IPCOM000234880D
Publication Date: 2014-Feb-12
Document File: 1 page(s) / 18K

Publishing Venue

The Prior Art Database


There are many situations where we reply to sent mail that relate to some mail thread in the same way but to different senders. As an example, an employee that is announced about his promotion to his all department is expected to get similar "congratulations" mails from his colleagues. Therefore, to be polite, that employee will be expected to send again and again the same "thank you" message to all his senders. Existing solutions for auto-reply require manual configuration of automatic response messages. Some auto-reply solutions allow to manually determine to which recipients to reply and on what condition (e.g., out-of-work status or mails related to project 'X'). Therefore, a user that wishes to automate his replies whenever they pop-up, needs to manually update her email client's auto-reply configuration again and again. We propose to automate the process, by actively discovering mail conversation patterns that may trigger an auto-reply message from the user's side. Hence, every time we detect such pattern we can provide the user an automatic suggestion (recommendation) for auto-replay configuration that covers the repeated reply pattern of the user.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 73% of the total text.

Page 01 of 1

Recommendation of auto -

The proposed solution has two main components.

1. Conversation repeat pattern detection component

2. Automatic auto-reply configuration recommender

Once component 1 detects the repeating pattern a signal is passed with metadata about the pattern to the recommendation component which generates an auto-reply configuration suggestion for the user's approval.

Once the user approves, the auto-reply feature of the email client takes care for replying to further senders.

We now shortly describe how the two components can be realized.

1. Detection component:

When a user replies to a given mail, the component looks for mails that were sent to the user with the same or similar topic/content.

The system then compares the similarity between the current user reply content to the content of previously replied mails.

If the similarity exceeds some threshold and/or the number of previously replied similar mails exceed some number, the component triggers a notification to the Recommendation component and includes a metadata about the detected patterns, such as the mails content, and user replies to previous mails and the current one.

2. Recommendation component:

The recommendation component takes the input from detection component and generates a recommendation for auto-reply configuration for the user to be added in response to user's approval.

The recommendation may auto generate the reply message using user's replies to previous mails, thought another...