Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
2000-03-23
2002-10-08
Mizrahi, Diane D. (Department: 2175)
Data processing: database and file management or data structures
Database design
Data structure types
C707S793000, C707S793000, C707S793000
Reexamination Certificate
active
06463433
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to computer database systems and more specifically to distributed computer database systems.
BACKGROUND OF THE INVENTION
As is generally recognized in the art, two of the most significant changes in the nature of information processing in the last decade are the transition from primarily alphanumeric text processing to multimedia processing and the connection of formerly isolated computers by networks, which have been connected in turn by intranets and the Internet. The first change has resulted in computer images becoming as common on computers as text. The second change has resulted in vast quantities of information, both text and multimedia, being accessible to individuals. This increase in information availability to individuals has come at the cost of increasing difficulty in finding relevant information.
a) Word Based Search Engines
Search engines have been developed to assist in information retrieval, but they are still primarily based on matching words in a query with words in text documents. In practice, this means that they cannot typically search effectively for features of images and other kinds of multimedia. Word based systems and non-word based systems presently employ separate and distinct approaches to extract relevant information.
One way of extracting information from a word based database is to submit an information request in the form of a query. Responsive to the query, a computer can extract information from the database that is related to that specified by the query. The extracted information can be used for determining the degree of “similarity” or “relevance” between a query and an object in the database. A variety of computer-implemented similarity measures have been developed for comparing a query with an object in the database when the query and database information are documents in a natural language. A commonly used measure of similarity is the cosine measure. The cosine measure is given by the formula, COS (v, w), where the vector v denotes the query and the vector w denotes the document. These vectors are in a space in which each possible word (or set of synonymous words) represents one dimension of the space. Further information regarding the cosine measure can be had with reference to G. Salton.
Automatic Text Processing
. Addison-Wesley, Reading, Mass., 1989; and G. Salton, J. Allen, and C. Buckley. “Automatic structuring and retrieval of large text files,”
Comm. ACM
, 37:97-108, 1994.
b) Non-Word Based Search Engines
As noted above, non-word based techniques currently employ approaches to extracting relevant information that are different and distinct from those used in word based systems. Non-word based information retrieval techniques are utilized advantageously, for example, in the field of medicine for extracting diagnostic information from images of the human body. Lung cancer is one of the most difficult cancers to cure. Early detection is important to improve the recovery rate. Chest CT scans are more effective than conventional chest X-ray techniques, but CT scans result in many more images to be examined, making computer assistance essential for mass screening programs. Computer aided diagnosis of CT images requires the extraction of a large number of features such as the lung area, blood vessels, air clusters and tumors. These features are detected using a computer-implemented thresholding algorithm along with smoothing to remove artifacts of the CT scanner. These features, in turn, have a complex structure involving attributes such as their shape, area, thickness and position within the lung. In implementing such algorithms on a computer for detecting such features, it is useful to employ an object database. An object database is a collection of data or information objects organized and stored on a computer storage medium pursuant to a data model. Each Information object has a type, such as, image, sound or video stream, as well as data object, e.g., text file or structured document. Each information object is identified uniquely by an object identifier (OID). An OID can be an Internet Universal Resource Locator (URL) or some other form of identifier such as a local object identifier. Databases containing images, sound and/or video streams can include not only the information objects themselves but also features and metadata. The data model used for such a database can support the representation of information at many levels of abstraction, including:
1. The data representation level, which contains the actual data of the information object.
2. The data object level, which stores data objects (such as lines and regions) extracted from the information object. The objects on this level do not have a domain interpretation.
3. The domain object level, which associates a domain object with each object at the data object level.
4. The domain event level, which associates domain objects with each other, providing the semantic representation of spatial or temporal relationships.
A feature at the data object level (i.e., at Level 2, above) can be represented as a set of domain-independent data such as lines and regions. A feature at the domain level (i.e., Levels 3 and 4, above) can be represented as a set of domain objects related to one another by domain-dependent relationships.
Consider another example in medicine. Mammography is one of the most effective methods of early detection of breast cancer, one of the leading causes of cancer among women. Manual reading of mammograms is labor intensive, so computer assistance is essential. A very large number of features in mammograms have been identified as being important for proper diagnosis, such as clustered microcalcifications, stellate lesions and tumors. Each of these can be represented as a set of medical domain objects with a complex structure. For example, a stellate lesion has a complex structure, consisting of a central mass surrounded by spicules. The spicules, in turn, have a complex, star-shaped structure. Extracting these complex domain objects and their relationships with each other is important for effective detection of breast cancer.
Features of images, sound and video streams can be represented in a computer system as a set of data structures stored in a database. The features can be categorized into the following types:
Features, such as the name of the photographer or date taken, that cannot be directly extracted from an information object, and are often descriptive of other data regarding the information object. Features of this kind are called metadata.
Features that can be extracted directly from the information object at the time of insertion into the database.
Features that are not calculated until needed.
Features can be as simple as the value of an attribute such as brightness of an image, but many features are more complicated and are thus represented using a complex data structure. An example of such a complicated feature is a representation of the structure of a stellate lesion in a mammogram.
Typically, features can be extracted from structured documents by parsing the document to produce data structures, and can be extracted from unstructured documents by using one of the many feature extraction algorithms that have been developed for implementation on a computer. As in the case of structured documents, feature extraction from an unstructured document produces data structures. A large variety of feature extraction algorithms have been developed for media such as sound, images and video streams. For a discussion of such algorithms, reference should be had to A. Del Bimbo, editor.
The Ninth International Conference on Image Analysis and Processing
, volume 1311. Springer, September 1997. For example, medical images typically use edge detection algorithms to extract the data objects, while domain-specific knowledge is used to classify the data objects as medically significant objects, such as blood vessels, lesions and tumors. Fourier and wavelet transformations as well as many filtering algorithms are also used
Jarg Corporation
Kudirka & Jobse LLP
Mizrahi Diane D.
Veillard Jacques
LandOfFree
Distributed computer database system and method for... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Distributed computer database system and method for..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Distributed computer database system and method for... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2949773