Abstract: Determining synonyms of words in a set of documents. Particularly, when provided with a word or phrase as input, in exemplary embodiments there is afforded the return of a predetermined number of “top” synonym words (or phrases) for an input word (or phrase) in a specific collection of text documents. Further, a user is able to provide ongoing and iterative positive or negative feedback on the returned synonym words, by manually accepting or rejecting such words as the process is underway.

Abstract: Methods and arrangements for executing an audio search. A search query in a first language variety is accepted, the search query being in a form of at least one of: text and audio. A corpus of material in the first language variety is accessed, and first language variety baseforms are obtained via data obtained from a second language variety. A first language variety phonetic model is built, and the first language variety phonetic model and the second language variety are employed in executing an audio search based on the accepted search query.

Abstract: Methods and arrangements for executing an audio search. A search query in a first language variety is accepted, the search query being in a form of at least one of: text and audio. A corpus of material in the first language variety is accessed, and first language variety baseforms are obtained via data obtained from a second language variety. A first language variety phonetic model is built, and the first language variety phonetic model and the second language variety are employed in executing an audio search based on the accepted search query.

Abstract: Techniques, systems, and articles of manufacture for determining the current logical state of a social media communication thread. A method includes computing an initial probability for applicability of each of multiple logical states for a first entry in a social media communication thread, wherein each logical state corresponds to a stage of interaction between customers of an enterprise and/or agents of the enterprise based on features derived from content of entries in the communication thread, network structure of entries, and identity of authors of entries, computing a transition probability between each subsequent consecutive entry in the communication thread, wherein the transition probability indicates the probability of moving from one logical state to another, and determining the current logical state of the communication thread based on the computed initial probability for the first entry and the computed transition probability between each subsequent entry in the communication thread.

Abstract: Techniques for evaluating one or more spoken language skills of a speaker are provided. The techniques include identifying one or more temporal locations of interest in a speech passage spoken by a speaker, computing one or more acoustic parameters, wherein the one or more acoustic parameters capture one or more properties of one or more acoustic-phonetic features of the one or more locations of interest, and combining the one or more acoustic parameters with an output of an automatic speech recognizer to modify an output of a spoken language skill evaluation.

Abstract: In accordance with an embodiment of the invention, there is provided a technique for permitting a machine to discover classes and topics that data contains and to annotate data objects with those identified classes. The technique enables machines to group and annotate data objects in ways that are meaningful and intuitive for a user of the data objects. An interactive method uses clustering, along with feedback from a user on the clustering output, to discover a set of classes. The feedback from the user is used to guide the clustering process in the later stages, which results in better and better discovery of classes and annotation with more and more human feedback. A method can be used to produce labeled data that involves discovering classes and annotating a given dataset with the discovered class labels. This is advantageous for building a classifier that has wide applications, such as call routing and intent discovery.

Abstract: A phonetic vocabulary for a speech recognition system is adapted to a particular speaker's pronunciation. A speaker can be attributed specific pronunciation styles, which can be identified from specific pronunciation examples. Consequently, a phonetic vocabulary can be reduced in size, which can improve recognition accuracy and recognition speed.

Abstract: A method and system for evaluating cross-domain clusterability upon a target domain and a source domain. Target clusterability is calculated as an average of a respective clusterability of at least one target data item comprised by the target domain. Target-side matchability is calculated as an average of a respective matchability of each target centroid of the target domain to source centroids of the source domain, wherein the source domain comprises at least one source data item. Source-side matchability is calculated as an average of a respective matchability of each source centroid of said source centroids to the target centroids. Source-target pair matchability is calculated as an average of the target-side matchability and the source-side matchability. Cross-domain clusterability between the target domain and the source domain is calculated as a linear combination of the calculated target clusterability and the calculated source-target pair matchability.

Abstract: A method, an apparatus and an article of manufacture for automatic speech recognition. The method includes obtaining at least one language model word and at least one rule-based grammar word, determining an acoustic similarity of at least one pair of language model word and rule-based grammar word, and increasing a transition cost to the at least one language model word based on the acoustic similarity of the at least one language model word with the at least one rule-based grammar word to generate a modified language model for automatic speech recognition.

Abstract: A process for evaluating cross-domain clusterability upon a target domain and a source domain. The cross-domain clusterability is calculated as a linear combination of a target clusterability and a source-target pair matchability, by use of a trade-off parameter that determines relative contribution of the target clusterability and the source-target pair matchability. The target clusterability quantifies how clusterable the target domain is. The source-target pair matchability is calculated as an average of a target-side matchability and a source-side matchability, which quantifies how well target centroids of the target domain are aligned with the source centroids and how well source centroids of the source domain are aligned with the target centroids, respectively.

Abstract: A computer system for evaluating cross-domain clusterability upon a target domain and a source domain. The cross-domain clusterability is calculated as a linear combination of a target clusterability and a source-target pair matchability, by use of a trade-off parameter that determines relative contribution of the target clusterability and the source-target pair matchability. The target clusterability quantifies how clusterable the target domain is. The source-target pair matchability is calculated as an average of a target-side matchability and a source-side matchability, which quantifies how well target centroids of the target domain are aligned with the source centroids and how well source centroids of the source domain are aligned with the target centroids, respectively.

Abstract: A computer program product evaluating cross-domain clusterability upon a target domain and a source domain. The cross-domain clusterability is calculated as a linear combination of a target clusterability and a source-target pair matchability, by use of a trade-off parameter that determines relative contribution of the target clusterability and the source-target pair matchability. The target clusterability quantifies how clusterable the target domain is. The source-target pair matchability is calculated as an average of a target-side matchability and a source-side matchability, which quantifies how well target centroids of the target domain are aligned with the source centroids and how well source centroids of the source domain are aligned with the target centroids, respectively.

Abstract: Techniques, an apparatus and an article of manufacture identifying one or more utterances that are likely to carry the intent of a speaker, from a conversation between two or more parties. A method includes obtaining an input of a set of utterances in chronological order from a conversation between two or more parties, computing an intent confidence value of each utterance by summing intent confidence scores from each of the constituent words of the utterance, wherein intent confidence scores capture each word's influence on the subsequent utterances in the conversation based on (i) the uniqueness of the word in the conversation and (ii) the number of times the word subsequently occurs in the conversation, and generating a ranked order of the utterances from highest to lowest intent confidence value, wherein the highest intent value corresponds to the utterance which is most likely to carry intent of the speaker.

Abstract: Described herein are methods, systems, apparatuses and products for reconstruction of a smooth speech signal from a stuttered speech signal. One aspect provides for accessing a stored speech signal having stuttering; identifying at least one stuttered region in the stored speech signal; modifying the at least one stuttered region in the stored speech signal; and responsive to modifying the at least one stuttered region, reconstructing a smooth speech signal corresponding to the stored speech signal. Other embodiments are disclosed.

Abstract: A system and associated method for cross-guided data clustering by aligning target clusters in a target domain to source clusters in a source domain. The cross-guided clustering process takes the target domain and the source domain as inputs. A common word attribute shared by both the target domain and the source domain is a pivot vocabulary, and all other words in both domains are a non-pivot vocabulary. The non-pivot vocabulary is projected onto the pivot vocabulary to improve measurement of similarity between data items. Source centroids representing clusters in the source domain are created and projected to the pivot vocabulary. Target centroids representing clusters in the target domain are initially created by conventional clustering method and then repetitively aligned to converge with the source centroids by use of a cross-domain similarity graph that measures a respective similarity of each target centroid to each source centroid.

Abstract: Methods and arrangements for employing descriptors for agent-customer interactions are disclosed. Filtering the pooled records based on one or more predetermined criteria is done such that analyzing the filtered records and comparing one interaction between an agent and a customer with another interaction between an agent and a customer may occur.

Abstract: Described herein are methods, systems, apparatuses and products for reconstruction of a smooth speech signal from a stuttered speech signal. One aspect provides for accessing a stored speech signal having stuttering; identifying at least one stuttered region in the stored speech signal; modifying the at least one stuttered region in the stored speech signal; and responsive to modifying the at least one stuttered region, reconstructing a smooth speech signal corresponding to the stored speech signal. Other embodiments are disclosed.

Abstract: A method, an apparatus and an article of manufacture for mapping authors across multiple forums. The method includes creating a database that contains publicly observable information pertaining to multiple authors from multiple forums, generating a mapping between at least a first one of the authors from one of the forums and at least a second one of the authors from another of the forums in the database based on a comparison of structured information, unstructured user generated content information and network information, and generating a score of mapping between the first and the second authors by considering a weighted sum of the number of times the structured information, the unstructured user generated content information and the network information match between the first and the second authors.

Abstract: Systems, methods, and apparatuses for providing a safety arrangement for an integrated heater, circulation pump, and motor for an appliance are provided. In an example safety arrangement, the circulation pump assembly includes a heater, a pump, and a motor integrated into a single unit. The circulation pump assembly defines a first end, a second opposite end, and a housing extending therebetween. The safety arrangement includes a hub configured to at least partially surround the first end of the circulation pump assembly and a first shield configured to at least partially surround the housing of the circulation pump assembly. The safety arrangement further includes a second shield configured to at least partially surround the housing of the circulation pump assembly. The hub, the first shield, and the second shield are configured to cooperate with one another to protect the circulation pump assembly from one or more hazardous conditions.

Abstract: Methods and arrangements for developing knowledge bases from social media. A question is obtained from social media. Social media are consulted, and a legitimacy of the question is ascertained. All the answers to the question are harvested from the social media including the rich media that is associated with these answers, and the question is filtered out if determined not to be legitimate.