Methods and apparatus for indexing and searching of multi-media web pages

[0001] This invention relates to electronic data storage, management and retrieval systems and more particularly to methods and apparatus for storing, indexing and searching data stored in and referenced by Web pages.

[0002] The Internet, and particularly the World Wide Web, allows multimedia information to be globally disseminated. Web pages expressed in a hypertext markup language often integrate information expressed in natural language text with static images, audio and video presentations and information generated by executing identified programs. While widely used “search engines” provide the ability to search for desired information based on the textual content of Web pages, there is a need for improved methods and apparatus for indexing and searching the multimedia content which is incorporated into Web pages.

[0003] The present invention takes the form of methods and apparatus for first analyzing each Web page to be indexed to identify media data which are incorporated by reference into that Web page; then extracting information describing the media data thus identified from the referencing Web page, from the media file itself, and from other sources; then inserting the extracted information as text annotations into a copy of the original Web page used for indexing purposes, and finally presenting the annotated Web page for processing by conventional text-based Internet indexing and searching facilities. The resulting index with store the association between the original Web page and the metadata which describes that page's media data content.

[0004] In accordance with the invention, a media-specific parsing program may be advantageously used to extract metadata already stored in external media files or other media resources referred to in a Web page in accordance with the format specifications for that referenced data's particular media type. A media content processing program is also preferably executed to analyze the media data to generate additional text-based information which characterizes the content of the referenced media data. In addition, a further program may be executed to acquire auxiliary data from one or more sources external to the media data being described, including such sources as the Internet, keyboarded descriptions entered by the user, or information describing the media data contained in system directories.

[0005] As contemplated by the invention, after the metadata describing the media data is obtained, it is combined to form a set of textual annotations in a standard text-based representation, preferably using Extended Markup Language (XML). These annotations are inserted into a copy of the original Web page which contained the references to the media data and the resulting annotated Web page is then indexed using conventional text-based indexing and search engines.

[0006] These and other objects, features and advantages of the present invention may be better understood by considering the following detailed description of the preferred embodiment of the invention. In the course of this description, reference will frequently be made to the attached drawings.

[0009] In the description of a specific embodiment of the invention that follows, the terms “media data” and “multimedia data” include digital image, video and audio data, and is to be distinguished from character-based text data which may be readily indexed and managed by conventional text processing mechanisms. Data which describes media data are interchangeably referred to as “metadata” and “annotations,” these terms being used interchangeably in the description which follows to identify a collection of indexable and/or manipulable attributes or properties expressed in natural language text (such at titles, media file attributes, file content descriptors, copyright notices, and the like).

[0010] The term “Web page” as used herein refers to an Internet addressable unit of data, such as a named file or data returned by an executable server program, which can be displayed by a Web browser program. The text data for Web pages are typically expressed in Hypertext Markup Language (HTML) but may also be expressed using the Structured Graphics Markup Language (SGML), or the Extended Markup Language (XML), all of which are character-based textual representations which may contain markup tags which identify non-text data, such as image, audio or video data, or program files. The markup tag typically contains the multimedia data's identifier, such as an Internet URL. Both Web pages and the media data which is incorporated by reference into the Web page are retrieved for presentation to a user from local storage using operating system file access routines, or from remote locations using a suitable request-response network communications protocol, such as the conventional HTTP/TCP-IP transmission mechanism used by the Internet World Wide Web facility.

[0011] HTML, in its preferred forms, as been defined in specifications which have continued to evolve to meet needs of users and developers. HTML 2.0 was developed under the aegis of the Internet Engineering Task Force (IETF) to codify common practice in late 1994 and is described in RFC 1866 (November, 1995). The efforts of the World Wide Web Consortium's HTML Working Group to codify common practice resulted in HTML 3.2 (January 1997). HTML 4.0, the latest version of which is currently available at the URL http://www.w3.org/TR/html40, extends HTML with mechanisms for style sheets, scripting, frames, embedding objects, and other enhancements.

[0012] The Extensible Markup Language (XML) is a subset of SGML which was designed to enable generic SGML to be served, received, and processed on the Web in the way that was previously possible with HTML. An XML document, as specified in the World Wide Web Consortium's Recommendation entitled “Extensible Markup Language (XML) 1.0” (February, 1998), may consist of one or many storage units called entities; all of which have content and which are typically identified by name. Each XML document has one entity called the document entity, which serves as the starting point for the XML processor and may contain the whole document. The XML specification permits an XML document to refer to one or more external entities by an appropriate identifier (URI) so that the content of the external entity referred to may be incorporated into the XML document. Entities may be either parsed or unparsed. An unparsed entity is a resource whose contents may or may not be text, and if text, may not be XML. Each unparsed entity has an associated notation, identified by name. Beyond a requirement that an XML processor make the identifiers for the entity and notation available to the application, XML places no constraints on the contents of unparsed entities, and XML documents may accordingly contain media data as unparsed data. XML documents may be translated into HTML using a suitable translator in accordance with cascaded style sheets (CSS) or the Extensible Style Language (XSL).

[0013] As used herein, the term “hypertext markup language” should accordingly be understood to include all of the evolving versions of HTML, as well as other character-based hypertext markup languages such as SGML and XML.

[0014] HTML's multimedia features allow authors to include images, applets (programs that are automatically downloaded and run on the user's machine), video clips, and other HTML documents in their pages. Commonly, in order to completely render a web page (i.e, to display all of the referenced text and images, as well as to play referenced sound, video or program files), it is necessary for the web browser program to scan the HTML text, identify the references to included resources that need to be fetched, and issue a sequence of separate requests using the Hypertext Transfer Protocol (HTTP) to obtain a current copy of each additional item of referenced data which may then be rendered by the browser or by a “helper” application capable of rendering data of a particular type. The rendering of an XML document may similarly require multiple HTTP request/response exchanges to assemble the entire document, including exchanges for fetching unparsed entities containing image, video, audio or program data which is rendered as part of the Web page.

Automatic Annotation of Web Pages

[0015] As contemplated by the present invention, Web pages are pre-processed to enrich them with text-based annotations which describe the multimedia data which is incorporated by reference into a copy of the original Web page which be used for indexing purposes. The index will maintain an association between the original Web page and the media-sensitive metadata. The additional metadata which is inserted into each Web page to describe its multimedia content may then be processed by conventional Web page indexing and searching software to allow multimedia data to be more readily located, presented to users, and otherwise processed.

[0016] The mechanism for automatically inserting searchable character-based annotations into a Web page which describe the multimedia component of that Web page is illustrated generally in FIG. 1 of the drawings. First, the Web page seen at 11 is automatically analyzed at 13 to identify the presence of markup tags which specify the URLs of external resources which supply multimedia content for the Web page 11.

[0017] The detection of one or more tags containing URLs which specify image data trigger processes which extract metadata from the identified image content as well as others sources as indicated 15. Similarly, markup tags containing URLs which identify audio or video data are processed as illustrated at 16 and 17 and respectively to extract metadata which describes each multimedia entity. As seen at 20, the extracted metadata is converted into annotations expressed in a character-based format suitable for processing by conventional Web page indexing and searching mechanisms. The annotations added to the copy 12 of the Web page 11 are preferably expressed in the Extended Markup Language (XML). These annotations are inserted at 22 into a copy 12 of the original Web page 11 to enhance its descriptive content before the copy 12 is indexed or published via the Internet 24. The information contained in the inserted annotations make that information available for indexing by existing search engines illustrated by the index server 25 in FIG. 1. Any authorized Internet user may employ a conventional Web browser 27 to communicate with the index server 25 to obtain the URL of Web page 11 by performing conventional keyword searches which employ search terms which characterize page 11's multimedia content. For example, a search might be conducted for Web pages which incorporate an audio rendition of “Stardust” or for Web pages that contain a JPEG image picturing a “dove” by using conventional search engines to identify web pages which contain the words “stardust” or “dove” respectively.

[0018] The mechanism for automatically annotating a Web page with metadata describing that pages multimedia components is illustrated in more detail by the flow chart of FIG. 2. The process is entered at 31 and the first Web page to be annotated is selected from a collection of such web pages (for example, from the contents of a directory folder containing Web pages to be published on the Internet, optionally further including all other Web pages specified in links within Web pages using a “crawler”). The Web pages being indexed may be stored locally or fetched via the Internet.

[0019] The process of scanning or parsing each Web page is initialized at 33 and proceeds at 34. As indicated at 35, the scanning process searches the selected Web page for markup tags which specify multimedia content. The identification of multimedia tags may be performed by an conventional HTML, SGML or XML parser. For example, the standard Java class DocumentParser in the package javax.swing.text.html.parser contained in the Java Platform 1.2 API Specification as promulgated by Sun Microsystems, Inc. 901 San Antonio Road, Palo Alto, Calif., 94303 may be used to parse HTML Web pages, and Oracle's XML Parser for Java v2, available from Oracle Corporation, 500 Oracle Parkway, Redwood Shores, Calif. 94065), may be used to identify multimedia tags in XML documents. The identification of multimedia tags may be performed as part of a concurrently performed process of validating the HTML, SGML, or XML pages (documents) prior to publication.

[0020] Alternatively, the text content of the web page may be directly scanned for the presence of tags which include the URLs of imbedded media files. For example, a case insensitive character scan may be performed for the specific HTML character strings which begin multimedia tags, such as: “<img” (images), “<a” (links to other web pages), “<form” (form handling programs), “<area” (image mapped links to other web pages), “<frame” (web pages loaded into frames), “<embed” (audio or video), “<sound” or <bgsound “(background audio). When each such tag is identified, the scanning process may then extract the URL placed in the tag in accordance with that tag's standard format. Note that, when the Web page includes an optional <base> tag containing a protocol and pathname, “local” URLs found within tags may be fully resolved by combining them with the separately specified base URL. Note also that the URL of the referenced media data may serve three separate purposes: it may be used to access the multimedia data from the identified resource so that additional descriptive metadata may be extracted for indexing and searching purpose, the filename extension in the URL may be used to determine the type of multimedia data to perform type-specific processing, and the URL itself constitutes metadata which may be useful for indexing and searching functions.

[0021] In addition to the URL, the markup tag which identifies multimedia may include additional information which describes how the media content is integrated into the web page. For example, “<img>” tags may also include optional parameters such as an “alt” parameter which specifies descriptive text to be displayed when, for some reason, the referenced image file cannot be rendered, and “height” and “width” parameters indicating the size of the image as displayed in the Web page. This descriptive information may be extracted to form part of the metadata about the referenced multimedia data which is later inserted into the Web page for indexing and searching as discussed later.

[0022] Still further information may be obtained from the file system directories which are maintained by the storage system which stores the multimedia data. Such directories typically contain time stamp information indicating when media data files were initially created and last modified. This information may be captured from the system directories and included as part of the metadata describing the multimedia data.

[0023] Finally, the content of the media data itself may contain information which can be expressed in text form as metadata. To capture such information, the type and format of the media data may be determined as indicated at 38 in FIG. 2 from the MIME type designation or a registered type designation associated with the filename extension in the URL, or by identifying format-identifying characteristics of the media data.

[0024] The media type when determined may be used to select a specific process at 40 for extracting descriptive information which from the content of the file. For example, common graphics file formats and the filename extension characters used to identify them are listed below:

[0025]

EXT.	DATA FORMAT
BIFF	XITE 3D file format
BMP	Microsoft Windows bitmap format
BW	SGI Black & White Image File Format
CGM	Computer Graphics Metafile
DRAW	Acorn's object-based vector image file format [Link]
DWG	AutoCAD drawings file format information [Link]
FAX	The Group 3 Facsimile standard
DCX	Format (graphics format for fax)
EPSF	Encapsulated Postscript Files
FIG	The FIG V3.1 file format (used by the xfig utility)
FITS	Flexible Image Transport System
GIF	Graphics Interchange Format
HDF	Hierarchical Data Format
ICC	Used for Kodak printer
IFF	Interchange Format
JPEG, JPG	JPEG File Interchange Format (V1.02)
MIFF	Machine Independent Format
NAP	The NAPLPS objected-oriented format
netCDF	The network Common Data Form
PIX	Used by SGI Alias|Wavefront products
PCX	Used by PC Paintbrush
PNG	Portable Network Graphics Specification
PBM	Enhanced Portable Bitmap toolkit
RLE	Utah Run Length Encoded Format
RAS	Sun Raster File Format
RGB/RGBA	SGI Colour Image File Format
SLD/SLB	AutoDesk Slide File Format
SLD	Slide File Format Specification
SLB	Slide Library File Format Specification
SPRITE	Acorn's bitmap format for their RISC OS
TGA	Targa File Format
TIFF	Tag Image File Format
VIFF	Used by the Khoros Visualisation package
X	The AVS Image Format
XBM	X BitMap Format
XPM	X PixMap Format
XWD	X Window Dump Format

Audio files, such as “.wav” Wave files and “.mid” MIDI files, and video files, such as “.mpg” MPEG compressed video, are similarly indicated by the filename extension MIME type.

[0026] Using the URL in the markup tag to fetch the media file, and using the filename extension media-type specifier in the URL to select a media-format-type specific routine to extract descriptive information from the content of the identified media data, that descriptive information may then be appended at 42 to the other annotations which describe the media data.

[0027] The type-specific media extraction of metadata embedded in the digital media is performed at 40 according to the file format specifications for the particular type of media file being processed. The type-specific file format specifications define the structure of the media data and indicate where metadata of interest is located within the media data, allowing it to be extracted at 40, transformed into a standard text-based format, preferably XML, and appended to the other annotations at 42. The MIME type of the media source may be used to select, dynamically load and execute an appropriate, type-specific parsing routine adapted to extract desired metadata from media data of the identified mimetype. A set of standard parsers for use with widely used media types may be extensibly augmented by additional, user-defined parsers which may be “plugged into” the framework at run-time, thereby extending the range of media formats handled by the system.

[0028] The metadata extracted from the content of the media data is appended at 42 to the metadata previously obtained from other sources, including the markup tags which identified the media data, from system directories, and from other sources, such as keyboarded input accepted from a human editor and supplied in response to automatically generated prompts generated during the course of the annotation process.

[0029] In accordance with the invention, the combined metadata describing each of the various multimedia resources which are incorporated into the Web page being scanned are represented in text (character-based) form and inserted into the Web page to enhance its content as seen at 46 in FIG. 2. These inserted text annotations may advantageously conform to both the XML specification and to the Resource Description Framework (RDF) Model and Syntax Specification, a World Wide Web Consortium (W3C) Recommendation (available at http://www.w3.org/TR/REC-rdf-syntax). The RDF Recommendation introduces a model for representing metadata as well as a syntax for encoding this metadata in a manner that maximizes the interoperability of independently developed Web servers and clients. RDF uses the Extensible Markup Language XML and specifies semantics for data based on XML in a standardized, interoperable manner.

[0030] The extracted metadata is preferably classified in accordance with predefined annotation element types or predefined attributes of an element type. For example, the attribute names listed below may be used for enhancing the content of Web pages in accordance with the present invention. These listed attributes are also used in the Oracle interMedia Annotator, as disclosed in U.S. patent application Ser. No. 09/410,781 filed on Oct. 1, 1999 by Alok Srivistava, Paul Lin and Marco Carrer, the disclosure of which is incorporated herein by reference. That prior application describes the use of metadata describing multimedia data as used in XML annotations which are stored in a relational database for indexing and searching multimedia resources also stored in the database. See also, the “Oracle8i interMedia Audio, Image, and Video User's Guide and Reference.” Release 8.1.,5 Oracle Corporation, part number A67299-01, (1999).

[0031]

Attribute	Description
Generic Media Annotations
MEDIATITLE	Title of the media
MEDIACOPYRIGHT	Copyright information of the
	media
EDIAPRODUCER	Producer of the media
MEDIADURATION	Duration (in seconds) of the
	media
MEDIACONTENT_DATE	Creation date of the media
	content
MEDIAMODIFICATION_TIME	Modification time of type
	Java.lang.Date
MEDIACREDITS	Credits for content providers
MEDIASIZE	Size of the media
MEDIAFORMAT_ENCODING	Format of the media
MEDIAUSER_DATA	String containing all user data
MEDIALANGUAGE	Language of the media
MEDIABITRATE	Bitrate of the media (in bits/
	second)
MEDIACATEGORY	Media category/genre
MEDIASOURCE_URL	Location/URL of the parsed
	media source
MEDIASOURCE_PROTOCOL	URL protocol of the media
	source
MEDIASOURCE_MIME_TYPE	MIME type of the media and
	its samples
MEDIASOURCE_DIRECTORY	Directory where the source
	is stored
MEDIASOURCE_FILENAME	Filename of the source
MEDIASOURCE_FILE_FORMAT	Media file format
MEDIAAUTHORING_TOOL	Software tool used to create
	the media
Audio Annotations
AUDIOAUDIO_ARTIST	Main artist for the audio clip
AUDIOAUDIO_BITS_PER_SAMPLE	Number of bits per sample
AUDIOAUDIO_SAMPLE_RATE	Audio sample rate (in
	samples/second)
AUDIOAUDIO_NUM_CHANNELS	Number of audio channels
Image and Video Annotations
VIDEOFRAMERATE	Video frame rate (in frames/
	second)
VIDEOFRAMESIZE	Video frame size (in bytes)
VIDEOSRCHEIGHT	Video height (in pixels)
VIDEOSRCWIDTH	Video width (in pixels)
VIDEOHORIZONTALRES	Horizontal resolution (in
	pixels/inch)
VIDEOVERTICALRES	Vertical resolution (in pixels/
	inch)
VIDEOISGRAYSCALE	Whether the video has colors
VIDEO_DEPTH	Number of bits for the color
	depth

[0032] While the above-noted attribute names and meanings may be used to particular advantage in those systems which employ like attribute names, such as the interMedia Text Engine, automated routines which generate annotations having different or additional attribute and element names may be used. The selection of a particular schema is made to best integrate the operation of the annotation-generating preprocessor with the operation of existing indexing and searching facilities.

[0033] When all of the Web pages in the collection have been enhanced with inserted annotations describing the included multimedia data, as determined at 48 in FIG. 2, the annotated Web page copies may then be indexed as indicated at 50 in conventional ways. Simply making such enhanced Web pages available on the Web allows them to be indexed by existing automated search engines (“Web crawlers” or “spiders”) such as those used by the indexes are publicly available at www.hotbot.com, www.altavista.com, www.excite.com, www.lycos.com, etc. Alternatively, the enhanced Web pages may be indexed for local use using conventional indexing mechanisms, and then discarded, leaving only the original Web page in storage, thereby conserving storage space and reducing Web page transport times. Note that, after indexing is completed, the URL of each indexed Web page which is supplied to users by the indexing or searching facilities should be the URL of an available Web page and not the URL of a discarded original or copy that is no longer available because it was discarded.

[0034] It is to be understood that the specific embodiment of the invention which has been described is merely illustrative of one application of the principles of the invention. Numerous modifications may be made to the system described without departing from the true spirit and scope of the invention.