Browse Prior Art Database

A method and system to optimize candidate phrases sequencing for Input Method Disclosure Number: IPCOM000248682D
Publication Date: 2016-Dec-27
Document File: 9 page(s) / 165K

Publishing Venue

The Prior Art Database


Modern people are spending more time to respond or comment to an article or an instant message, a variety of input methods were built into the mainstream input method software which help users to input characters quickly especially for Chinese. These software can leverage both local and network lexicons and sorting techniques to improve user experience, but there are loads of occasions the expected phrases are not presented as the first phrase or in the first page thus the user need to navigate the long candidates list to select the desired phrases/characters, which significantly reduces the user input speed. This disclosure is to introduce a method and system to optimize candidate phrases sequencing for Input Method.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 53% of the total text.


A method and system to optimize candidate phrases sequencing for Input Method

Claim Points: ● A method to identify the context and scene for the text content the user is responding to ● A method to optimize the sequencing of the candidate phrases based on the identified background attributes

Application Classifications:


High Level Process:


Once an application is launched by an user, it’s classification will be detected immediately in the background. An application 1. can be but not limited to, a webpage, a desktop application, a mobile app.

For online forum or chatrooms, the corresponding industries will be identified and mapped to specific dictionaries either local or 2. remote, which then be assigned the highest priority value among all available dictionaries.

For microblog or instant messenger applications, it will go directly to semantic analysis of the text content of article or comments 3. depends on what the user is responding to. Semantic analysis is applied for both forum/chatroom and microblog/IM.

Given it’s an article, semantic analysis will be applied to the full context of the article and keywords to be extracted. 4.

While if the user is replying to a specific comment or message , then additionally the content of the comment/message will be 5.


analyzed that results in the next level keywords with higher priority than article keywords.

With the generated keywords, the industries involved will be detected again and the industries dictionaries priorities will be 6. adjusted.

For short articles/comments/messages, the properties of those keywords will be calculated and marked. 7.

When the user starts typing with input method software, the priority method mentioned above will be applied. Additionally, the 8. calculated properties(if any) will be taken in to account for the calculation of the final priority values of potential phrases .

While, if the application classification can not be identified , the existing priority/sorting mechanism of input method will take 9. effect.

Prioritization of Candidate Phrases:


Identify Industry as Site Level: For chatrooms and forums, it’s straight forward to get the industry involved by querying websites directory. It’s possible that the main website belongs to industry A but the subsite/module belongs to industry B. If that’s the case, the industry will be take as B which is the lowest level subsite/module. Thus, the associate dictionaries will be identified. For example, if the website is marked

as for Java programming, then candidate phrases within Computer Dictionary will get the highest site level priority value.

PI(Di) is the site level priority value for a candidate phrase Di.


Semantic Analysis of the Article: Upon semantic analysis on an article, the keywords will be generated, followed by the associated attributes of those keywords. For a candidate phrase, the article level priority value is the sum of correlation value to each keywords: