Abstract: A method and apparatus are provided for improving the precision of information retrieval systems that use logical form searching techniques. Under one embodiment of the invention, several logical form triples are produced from the user's query and are combined together by restrictive logical operators to generate a compound logical form query. A search is then performed to find documents that meet the requirements set by the compound logical form query. In other embodiments, results generated by a logical form search are intersected with results from a word search to form a more precise set of results. In further embodiments of the invention, three pairs of search results are intersected with each other to form three sets of final results. These final results are then ranked based on the techniques used to form their constituent result pairs.

Abstract: The present invention provides a dialog system in which the subsystems are integrated under a single technology model. In particular, each of the sub-systems uses stochastic modeling to identify a probability for its respective output. The combined probabilities identify a most probable action to be taken by the dialog system given the latest input from the user and the past dialog states. An additional aspect of the present invention is an embodiment in which the sub-systems communicate with one another through XML pages, thus allowing the sub-systems to be distributed across a distributed network.

Abstract: Cluster- and pruning-based language model compression is disclosed. In one embodiment, a language model is first clustered, such as by using predictive clustering. The language model after clustering has a larger size than it did before clustering. The language model is then pruned, such as by using entropy-based techniques, such as Rosenfeld pruning, or by using Stolcke pruning or count-cutoff techniques. In one particular embodiment, a word language model is first predictively clustered by a technique described as P(Z|xy)×P(z|xyZ), where a lower-case letter refers to a word, and an upper-cluster letter refers to a cluster in which the word resides.

Abstract: A method and apparatus determine the likelihood of a sequence of words based in part on a segment model. The segment model includes trajectory expressions formed as the product of a polynomial matrix and a generation matrix. The likelihood of the sequence of words is based in part on a segment probability derived by subtracting the trajectory expressions from a feature vector matrix that contains a sequence of feature vectors for a segment of speech. Aspects of the method and apparatus also include training the segment model using such a segment probability.

Abstract: A method and data storage device are provided that reduce off-track motion due to resonant modes in the storage device. The off-track motion is reduced by sensing movement with a sensor located on a suspension assembly of the storage device. Based on this sensed movement, the sensor generates a signal that is used to drive a control system to minimize the off-track motion.

Abstract: A method and apparatus are provided that receive a caller identification packet from a central office. The caller identification packet is decoded to produce caller identification information, which is used to identify an audio message. Before presenting the audio message to the user, the method and apparatus disables ringing of the phone system. After the audio message has been presented to the user, ringing is re-enabled.

Abstract: A method and a computer-readable medium are provided for generating an interface in a personal computer system. The method includes is playing a primary view of a three-dimensional environment from the egocentric perspective of a virtual person in the environment. A glancing function is then performed, which provides animated movement of the virtual person's head to reveal a three-dimensional tool space that is normally hidden from view. In one embodiment, the three-dimensional tool space moves with or is linked to the virtual person's body so that it travels through the environment with the virtual person's body. In other embodiments, the tool space is associated with an object in the three-dimensional environment.

Abstract: A system for enhancing searching of a network has an architecture that supports a network indexing component capable of indexing terms located in pages on the network and capable of receiving a search query and returning a topic for each term in the query that matches an indexed term. An embodiment of the system further provides for a search tool capable of displaying text associated with assisting a user to perform a search of the Internet and an animation component capable of displaying an animated character. Further embodiments support a search tool component capable of soliciting a search query from a user and of invoking an instance of an Internet browser in which to display search results. A further embodiment of the invention providing at least one search option suggesting changing at least one search parameter and at least one option allowing the user to indicate that their search was successful.

Abstract: A method and apparatus for compensating for repeatable run-out errors in a disc drive is disclosed. The compensation is created by determining transfer function values for a servo loop in the drive as well as a sequence of repeatable run-out values for a portion of a disc. The repeatable run-out values are divided by respective transfer function values to produce compensation values that are inverse transformed to produce time-domain compensation values. These time-domain compensation values are injected into the servo loop to compensate for repeatable run-out errors.

Abstract: An optical head for performing optical data operations relative to a medium includes a slider body and at least one lens. Positioned between the slider body and the lens is an aperture stop for blocking the transmittance of light. The aperture stop includes an opaque layer that circumscribes a transparent region.

Abstract: A method and apparatus are provided for storing and retrieving data in a non-volatile manner. The data is stored in a magnetic cell having a magnetic moment with a direction. A conductor is positioned near the cell and carries a current formed by a current driver. A data detector detects a value for data stored in the cell based on a level of current driven through the conductor.

Abstract: A model is provided for formants found in human speech. Under one aspect of the invention, the model is used to synthesize speech. Under this aspect of the invention, the formant model is used to identify a most likely formant track for the synthesized speech. Based on this track, a series of resonators are used to introduce the formants into the speech signal.

Abstract: A method and apparatus are provided that reduce the training time associated with machine learning systems whose training time is proportional to the number of outputs being trained. Under embodiments of the invention, the number of outputs to be trained is reduced by dividing the objects to be modeled into classes. This produces at least two sets of model parameters. At least one set describes some aspect of the classes given some context, and at least one other set of parameters describes some aspect of the objects given a class and the context. Thus, instead of training a system with a large number of outputs, corresponding to all of the objects, the present invention trains at least two models, each of which has a much smaller number of outputs.

Abstract: There is disclosed a method of converting a bitmap to an object-based embroidery pattern by generating a skeleton from the bitmap and traversing paths and nodes identified in the skeleton, the embroidery pattern objects being generated during the traversal. The outline of the bitmap is used to define parts of the boundaries of the generated objects, which are laid down using a linear stitch type on the first traversal of a skeleton path and a fill stitch type on the second traversal of the skeleton path.

Abstract: A method and apparatus logs outgoing phone symbols dialed from an extension. To log the phone numbers, a switching network is coupled to the phone lines and passes a Dual Tone Multiple Frequency signal from the extension to a CODEC. The CODEC converts the analog signal it receives into a series of digital values that are passed to a digital signal processor. The digital signal processor decodes the series of digital values into a set of dialed-symbol values indicative of symbols dialed by the extension. The dialed-symbol values are packaged by a processing unit interface and transmitted to the processing unit where they are stored in a log database.

Abstract: A method and storage device are provided for initializing a polynomial linearizer in the storage device. The linearizer is initialized by identifying the coefficients of the linearizer polynomial. To reduce the computational intensity of this process, orthogonal-type coefficients for at least two orthogonal polynomials are identified, where the linearizer polynomial is formed as the sum of the orthogonal polynomials. The orthogonal-type coefficients are then combined to identify linearizer coefficients for the linearizer polynomial.

Abstract: A language model structure for use in a speech recognition system employs a tree-structured network model. The language model is structured such that identifiers associated with each word and contained therein are arranged such that each node of the network model with which the language model is associated spans a continuous range of identifiers. A method of transferring tokens through a tree-structured network in a speech recognition process is also provided.

Abstract: The present invention provides a user interface for a pen-based device that allows users to activate the full “card” for an entry using a single tap while also giving the user the ability to reach a menu of context dependent commands using a simple gesture. In particular, the present invention recognizes certain stylus and/or button inputs as context menu gestures. When these gestures are made, a context menu appears. The user is then able to select a command in the context menu that is to be performed. The present invention also allows a user to open context menus while only using one hand to operate and hold the device.

Abstract: A method and apparatus is provided for identifying patterns from a series of feature vectors representing a time-varying signal. The method and apparatus use both a frame-based model and a segment model in a unified framework. The frame-based model determines the probability of an individual feature vector given a frame state. The segment model determines the probability of sub-sequences of feature vectors given a single segment state. The probabilities from the frame-based model and the segment model are then combined to form a single path score that is indicative of the probability of a sequence of patterns. Another aspect of the invention is the use of a frame-based model and a segment model to segment feature vectors during model training. Under this aspect of the invention, the frame-based model and the segment model are used together to identify probabilities associated with different segmentations.

Abstract: A method and apparatus are provided for reading from a storage medium to form data values. A signal is generated from a sector on the storage medium and a section of that signal is identified as having a change in amplitude. A gain multiplier is activated to multiply that section of the signal by a gain value. The gain multiplier is deactivated at the end of the section so that the remaining portion of the sector signal is not multiplied by the gain value. This results in a defect adjusted signal that is applied to a detector to detect data values.