Language input architecture for converting one text form to another text form with tolerance to spelling, typographical, and conversion errors

• US 7,302,640 B2
• Filed: 10/21/2004
• Issued: 11/27/2007
• Est. Priority Date: 11/05/1999
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Chen, Zheng, Han, Jian, Lee, Kai-Fu
• Primary Examiner(s): Hutton; Doug
• Assistant Examiner(s): Tran; Quoc A.
• Application Number:
  US10/970,438
• Publication Number:
  US 20050086590A1
• Time in Patent Office: 1,132 Days
• Field of Search: 715/500, 715/533, 715/535, 704/9, 341/22, 455/566
• US Class Current: 715/257
• CPC Class Codes: G06F 40/129 Handling non-Latin characters, e.g. kana-to-kanji conversion
• CPC Class Codes: G06F 40/216 using statistical methods
• CPC Class Codes: G06F 40/232 Orthographic correction, e.g. spell checking or vowelisation
• CPC Class Codes: G06F 40/44 Statistical methods, e.g. probability models
• CPC Class Codes: G06F 40/53 Processing of non-Latin text kana-to-kanji conversion G06F40/129; vowelisation G06F40/232

Abstract

A language input architecture converts input strings of phonetic text to an output string of language text. The language input architecture has a search engine, typing models, a language model, and one or more lexicons for different languages. Each typing model is trained on real data, and learns probabilities of typing errors. The typing model is configured to generate a list of probable typing candidates that may be substituted for the input string based on probabilities of how likely each of the candidate strings was incorrectly entered as the input string. The language model provides probable conversion strings for each of the typing candidates based on probabilities of how likely a probable conversion output string represents the candidate string. The search engine combines the probabilities of the typing and language models to find the most probable conversion string that represents a converted form of the input string.

Claims

1. A method facilitating text form conversion comprising:

constructing one or more typing model(s);

training the one or more typing model(s) to determine probabilities that a user intended to enter an intended string when an input string was entered, wherein the one or more typing models are trained using more than one language, wherein entry of the input string is facilitated via modeless entry and comprises more than one language, wherein modeless entry facilitates automatically distinguishing between words of different languages, wherein text elements of the input string are divided into syllables in each of the trained languages and equivalently pronounced text elements of the input stripg are mapped to the same syllables, then the syllables are mapped to characters, and wherein the training is based on data collected from multiple users entering at least one training text; and

storing, on a storage media, the one or more typing model(s) after the training.

2. A method as recited in claim 1, wherein the training comprises:

maintaining a frequency count of the text elements of the input string mapping onto each of the syllables; and

determining probabilities that the syllables represent correct entry of the input string based on the frequency counts of the text elements of the input string.

3. One or more computer-readable media having computer-executable instructions that, when executed on a processor, direct a computer to perfonn the method as recited in claim 1.

4. A method of training one or more typing model(s), comprising:

reading a text string having multiple letters input by a user via a modeless entry, wherein modeless entry facilitates automatically distinguishing between words of different languages;

mapping syllables from more than one language to corresponding groups of letters in the text string;

mapping equivalently pronounced groups of letters in the text string to the same syllables;

maintaining a frequency count of the groups of letters in the text string mapping onto each of the syllables in each of the more than one languages;

determining probabilities that the syllables represent correct entry of the text string based on the frequency counts of the group of letters in the text string;

mapping the syllables to characters; and

storing, on a storage media, the one or more typing model(s) after training.

5. A method as recited in claim 4, wherein the text string comprises phonetic text.

6. A method as recited in claim 4, wherein the text string comprises a mixture of phonetic text and non-phonetic text.

7. One or more computer-readable media having computer-executable instructions that, when executed on a processor, direct a computer to perform the method as recited in claim 4.