Abstract: Systems and techniques are provided for receiving an input column and a search keyword and providing one or more suggested columns with which to merge the input column. A coverage score and a refinity score are calculated for potential columns based on the input column as well as a search score based on the search keyword. The one or more suggested columns may be determined based on the coverage score, refinity score, and/or the search score. The input column and/or a potential column may be modified based on a function and the modification may result in a plurality of modified input and/or potential columns. Coverage, refinity, and search scores may be calculated based on the modified columns.

Abstract: A communication device capable of determining an emotional or physiological state of a user of the device includes an input device that provides a signal indicative of one of an emotional and a physiological state of a user of the device. A controller receives the signal from the input device and interprets the signal into state data identifying one of an emotional and a physiological state of a user of the device. The device makes the state data available to be downloaded such that a device sending a communication can appropriately adapt its communication method and transport mode of user interface to suit the recipient's condition.

Abstract: A computer-implemented method for analyzing documents to discover noncompliance with an established norm is provided. The method can include receiving one or more terms indicating possible noncompliance with a pre-established norm, and, based upon the at least one term, constructing at least one grammatical unit. The grammatical unit can specify a predetermined syntax and can correspond to semantic content that is indicative of noncompliance with the pre-established norm, wherein the norm can include a statute, regulation, policy, or other standard. The method can further include identifying from among multiple electronic documents each document that contains one or more grammatical units specifying a predetermined syntax and corresponding to semantic content indicative of noncompliance with the pre-established norm.

Abstract: A structure and method for entering data into an electronic device (110). A deterministic data input element (130) coupled to the electronic device (110) is operable to accept a user-selected data input. Modulation of a motion of a non-deterministic data input element (140) coupled to the electronic device (11) is operable to select a varying precision of data input choices. A feedback element (120) coupled to one or more of the deterministic data input element (130) and the non-deterministic data input element (140) presents the data input choices to the user. The user enters data into electronic device (110) using one of the deterministic input element (130) and the non-deterministic input element (140) and the feedback element (120) displays a sequence of data choices to the user. The user then navigates a tree of possible data completions in order to select a particular data completion to be entered into the electronic device (110).

Abstract: A method (100) of crawling the Web (620) is disclosed. The method (100) crawls (120) Web pages on the Web starting from a given (110) set of seed Universal Resource Locators (URLs). Crawled Web pages are partitioned (140) into sets of relevant and irrelevant pages. A set of exclusion and/or inclusion patterns are discovered (150) from the sets of relevant and irrelevant pages, and subsequent crawling of the Web is restricted through the set of exclusion and/or inclusion patterns.

Abstract: A system, method and program storage device implementing a method for modeling a data generating process, wherein the modeling comprises observing a data sequence comprising irregularly sampled data, obtaining an observation sequence based on the observed data sequence, assigning a time index sequence to the data sequence, obtaining a hidden state sequence of the data sequence, and decoding the data sequence based on a combination of the time index sequence and the hidden state sequence to model the data sequence. The method further comprises assigning a probability distribution over time stamp values of the observation sequence, wherein the decoding comprises using a Hidden Markov Model. The method further comprises using an expectation maximization methodology to learn the Hidden Markov Model.

Abstract: In a speech recognition system, the combination of a log-linear model with a multitude of speech features is provided to recognize unknown speech utterances. The speech recognition system models the posterior probability of linguistic units relevant to speech recognition using a log-linear model. The posterior model captures the probability of the linguistic unit given the observed speech features and the parameters of the posterior model. The posterior model may be determined using the probability of the word sequence hypotheses given a multitude of speech features. Log-linear models are used with features derived from sparse or incomplete data. The speech features that are utilized may include asynchronous, overlapping, and statistically non-independent speech features. Not all features used in training need to appear in testing/recognition.

Abstract: A structure and method for entering data into an electronic device (110). A deterministic data input element (130) coupled to the electronic device (110) is operable to accept a user-selected data input. Modulation of a motion of a non-deterministic data input element (140) coupled to the electronic device (11) is operable to select a varying precision of data input choices. A feedback element (120) coupled to one or more of the deterministic data input element (130) and the non-deterministic data input element (140) presents the data input choices to the user. The user enters data into electronic device (110) using one of the deterministic input element (130) and the non-deterministic input element (140) and the feedback element (120) displays a sequence of data choices to the user. The user then navigates a tree of possible data completions in order to select a particular data completion to be entered into the electronic device (110).

Abstract: A communication device capable of determining an emotional or physiological state of a user of the device includes an input device that provides a signal indicative of one of an emotional and a physiological state of a user of the device. A controller receives the signal from the input device and interprets the signal into state data identifying one of an emotional and a physiological state of a user of the device. The device makes the state data available to be downloaded such that a device sending a communication can appropriately adapt its communication method and transport mode of user interface to suit the recipient's condition.

Abstract: A computer operating system (30) receives information from first and second application programs relating to control words for controlling the application programs. This information is contained in vocabulary A (74) and vocabulary B (76). The speech recognition tool (36) is arranged to accept the information from the first and second application programs and perform a comparison with digitized speech received at an input (60). A selected one of the application programs (32 and 34) is controlled according to the digitized speech received.

Abstract: A data entry apparatus (10) receives a first character key from a keypad (12). A set of characters associated with the first character key is displayed. A second character key from the keypad (12) is received, wherein a second set of characters is associated with the second character key. A character from the first set of characters is combined with a character from the second set of characters. A set of alternative n-grams is displayed, derived from the step of combining, in descending order based on a probability of frequency of use in a given language. When an alternative n-gram is detected as being actively highlighted, the n-gram is locked to create a locked n-gram and characters associated with subsequent character keys vary in their combinations with respect to the locked n-gram.

Abstract: A method and apparatus for generating a context dependent phoneme network as an intermediate step of encoding speech information. The context dependent phoneme network is generated from speech in a phoneme network generator (48) associated with an operating system (44). The context dependent phoneme network is then transmitted to a first application (52).

Abstract: A microprocessor 11 receives an ambiguous input from a key 2-9 and disambiguates the ambiguous input using a dictionary. The dictionary is stored in an efficient manner by being partitioned according to the first three letters of a word. A sequence of three letters in the dictionary is stored as eight bits of data, the first three letters of a word being stored using three pairs of bits, such that each pair of bits represents one of no more than four alternative letters.

Abstract: A method of text entry into a device by activating keys of a keypad (16), where a key represents various characters. Input is delivered to a processor (30) corresponding to the keys activated. A dictionary (33) is searched for candidate combinations of characters corresponding to the keys activated. The candidate combinations are rank ordered, for example using language model data and feedback (14, 38) is provided to a user indicating at least a highest rank ordered candidate combination.