Apple Files Patent Relating to a Video Game & Movie Database
On September 13, the US Patent & Trademark Office published Apple’s patent application titled Fuzzy string matching of media meta-data . While Apple’s patent generally relates to databases of media metadata, what we learn from this patent in earnest is that Apple could be planning to introduce a new media database relating to a video game and movie service. Apple’s patent specifically relates to the All Media Guide of Ann Arbor, Michigan as an example of a planned database. Such databases are available for access over the internet, for example, to populate records for media contained in a media library of a client computer. The All Media Guide includes allgame and allmovie. Considering that Apple’s QuickTime Guide has included movies and games for years, it would suggest that Apple’s original 2006 patent filing is relating to something new.
Key Patent Points
Apple states that within the patent’s description, “the term “database” is not meant to imply any particular structure or method of access. Rather, the term “database” is used expansively to broadly indicate a collection of data organized in some systematic manner. For example, the database may be merely organized as a table, or even some simpler structure, or the database may be as complex as a relational database or some other complex structure. Typically, the database is computer accessible and the methods described herein are computer-implemented methods.”
In Apple’s patent FIG. 2 they illustrate an example of a record 108 in the first media metadata database. By way of example, and not by way of limitation, the example record 108 is for holding metadata regarding a song. The record 108 includes, for example, a title field 252, a composer field 254, a performer field 256 and a Universal Product Code field 258. The entire contents of each field may be considered a token and, in addition, portions of each field may each themselves be considered a token. For example, taking the field 252 in FIG. 2, the following may be tokens corresponding to the field 252: “Under My Thumb” 252; “Under” 252- 1; “My” 252-2; and “Thumb” 252-3. While FIG. 2 illustrates a record 108 in the first media metadata database, records 104 in the second media metadata database 102 are generally similar in structure to the records 108 in the first media metadata database.
The Fuzzy Factor: The Rooling Stones
In one possible approach (not the invention) for matching the records of the second media metadata database to the records 108 of the first media metadata database, each of one or more records of the second media metadata database 102 is considered in turn. Specifically, the metadata in each record 104 of the second media metadata database 102 is, in turn, compared to metadata in each record 108 of the first media metadata database 106. Realizing that an exact match might not be found, a “fuzzy” match may be attempted, in an attempt to match words or other data that are similar, but not identical. However, fuzzy matches are expensive to compute for a single record, let alone for a number of records on the order of a million in each database to be matched.
In Apple’s FIG. 3 they illustrate a simplistic example of a table 201 that may be used to hold the results of the step 402 mapping, using the example of the first media metadata database 106. Rows 203a, 203b, 203c, etc. of the table 201 correspond to each token processed from the records 108a, 108b, etc. of the first media metadata database 106. The column 202 of a particular row (203, generically) holds an indication of the token processed. The column 206 holds one or more identifiers of records 108 in the first media metadata database 106 in which the token is found to appear. Thus, for each record of the first media metadata database 106 processed, if a particular token for a row 203 in the table 201 appears in that record 108, the identifier of that record 108 is added to the column 206 of the row 203 for that token. If a particular token appears in that record 108 but does not yet have a row 203 in the table, then a row 203 is created in the table 203 for the particular token, with the identifier for the record 108, in which the particular token appears, in the column 206 of the row 203 for the particular token.
The column 207 indicates the “popularity” of the token for the row 203. Each time the token for the row 203 appears, the popularity value in column 207 for that row 203 is incremented. One use for the popularity value is discussed later.
Thus, in one example, after processing the record 108 according to step 402 of the FIG. 4 flowchart, the column 202 of the table 201 would include the tokens “Under My Thumb”; “Under”; “My”; “Thumb”; “Mick Jagger”; “Mick”; “Jagger”; “Keith Richards”; “Keith”; “Richards”; “The Rolling Stones”; “The”; “Rolling”; “Stones”; and “9999999999″. For simplicity of example, the FIG. 3 table includes only the tokens “The” (row 203a), “Rolling” (row 203b), “Stones” (row 203c), and “The Rolling Stones” (row 203d). If, for example, a word is misspelled, the token would also be misspelled. For example, “Rooling” (a misspelling of “Rolling”) would have a separate entry in the column 202 from “Rolling.” Assuming a record identifier of the record 108 (FIG. 2) is “12345,” the corresponding entries in the column 206 would all include the value “12345.”
Apple’s Summary
The invention is, in one aspect, a method of matching a particular media metadata record to one of a plurality of records of a first media metadata database. For the particular media metadata record, for each of at least some of the tokens of the particular media metadata record, a table is processed that correlates that token to records of the first media metadata database. Based on the processing, it is determined in which of the plurality of records of the first media metadata database that token of the particular record appears. The determined records are match candidate records of the first media metadata database.
For each of at least some of the match candidate records of the first media metadata database, a value is determined for each match candidate record that indicates a degree of match between that match candidate record and the particular media metadata record. The determined values are processed to determine one of the match candidate records is a nominal match to the particular media metadata record.
In one example, the table that correlates tokens to records of the first media metadata database is generated by processing the plurality of records of the first media metadata database to correlate, to each of at least a subset of the tokens appearing in the metadata of the first media metadata database, the records in which that token appears.
In summary, then, in accordance with a broad aspect, the first media metadata database (to which the records of a second media metadata database are to be matched) is “preprocessed” such that tokens of the records of the first media metadata database are matched to record identifiers of records in which those tokens appear. To match a particular record of the second media metadata database, the tokens of that particular record are located in the table. Match candidate records in the first media metadata database include those records of the first media metadata database in which the tokens of the particular record appear. The match candidate record exhibiting the best match, if any, to the particular record is deemed to be the matching record.
Apple lists Mark Miller (San Francisco, CA) as the sole inventor of this patent. For full details of this patent, click here.
NOTICE: MacNN presents only a brief summary of patents with associated graphic(s) for journalistic news purposes as each such patent application and/or grant is revealed by the U.S. Patent & Trade Office. Readers are cautioned that the full text of any patent applications and/or grants should be read in its entirety for further details.
Written and researched by Neo.
