Patents by Inventor Imed Zitouni

Imed Zitouni has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).

  • Publication number: 20190034780
    Abstract: Described herein is a conversation engine that can be used in a system such as a personal digital assistant or search engine that combines a dynamic knowledge graph built during execution of a request and one or more static knowledge graphs holding long term knowledge. The conversation engine comprises a state tracker that holds the dynamic knowledge graph representing the current state of the conversation, a policy engine that selects entities in the dynamic knowledge graph and executes actions provided by those entities to move the state of the conversation toward completion, and a knowledge graph search engine to search the static knowledge graph(s). The conversation is completed by building the dynamic knowledge graph over multiple rounds and chaining together operations that build toward completion of the conversation. Completion of the conversation results in completion of a request by a user.
    Type: Application
    Filed: July 31, 2017
    Publication date: January 31, 2019
    Inventors: Marius Alexandru Marin, Paul Anthony Crook, Vipul Agarwal, Imed Zitouni
  • Publication number: 20190034795
    Abstract: Systems and methods for adding a new domain to a natural langue understanding system to form an updated language understanding system with multiple domain experts is provided. More specifically, the systems and methods are able to add a new domain utilizing data from one or more of the domains already present on the natural language understanding system while keeping the new domain and the already present domains separate from each other.
    Type: Application
    Filed: July 28, 2017
    Publication date: January 31, 2019
    Applicant: Microsoft Technology Licensing, LLC
    Inventors: Imed ZITOUNI, Dongchan KIM, Young-Bum KIM
  • Publication number: 20190035387
    Abstract: Described herein is a mechanism to adapt a machine learning model used in a language understanding model that has been trained using a first set of user input having a first set of features to effectively operate using user input having a second set of features. Losses are defined based on the first set of features, the second set of features or features common to both the first set and second set. The losses comprise one or more of a source side tagging loss, a reconstruction loss, an adversarial domain classification loss, a non-adversarial domain classification loss, an orthogonality loss, and target side tagging loss. The losses are jointly minimized using a gradient descent method and the resulting coefficients are used to retrain the machine learning model.
    Type: Application
    Filed: July 27, 2017
    Publication date: January 31, 2019
    Inventors: Imed Zitouni, Dongchan Kim, Young-Bum Kim
  • Publication number: 20180341928
    Abstract: Systems and methods are disclosed for task identification and tracking using shared conversational context. In one implementation, a first communication from a first user is received within a communication session. The first communication is processed to identify a first content element within the first communication. A second communication is received within the communication session. The second communication is processed to identify a second content element within the second communication. The first content element is associated with the second content element. Based on an association between the first content element and the second content element, a task is identified. An action is initiated with respect to the task.
    Type: Application
    Filed: May 25, 2017
    Publication date: November 29, 2018
    Inventors: Omar Zia Khan, Vipul Agarwal, Imed Zitouni
  • Publication number: 20180255005
    Abstract: Systems and methods are disclosed for personalized communications using semantic memory. In one implementation, a first communication is received from a user and processed to identify a first content element within the communication. The first content element is associated with a second content element within a content repository. A second communication that includes the first content element is received from the user. Based on an association between the first content element and the second content element within the content repository, a third communication that includes the second content element is generated and provided to the user in response to the second communication.
    Type: Application
    Filed: March 5, 2017
    Publication date: September 6, 2018
    Inventors: Vipul Agarwal, Omar Zia Khan, Imed Zitouni, Hisami Suzuki
  • Publication number: 20180254035
    Abstract: Automatic speech recognition systems can benefit from cues in user voice such as hyperarticulation. Traditional approaches typically attempt to define and detect an absolute state of hyperarticulation, which is very difficult, especially on short voice queries. This disclosure provides for an approach for hyperarticulation detection using pair-wise comparisons and on a real-world speech recognition system. The disclosed approach uses delta features extracted from a pair of repetitive user utterances. The improvements provided by the disclosed systems and methods include improvements in word error rate by using hyperarticulation information as a feature in a second pass N-best hypotheses rescoring setup.
    Type: Application
    Filed: June 15, 2017
    Publication date: September 6, 2018
    Inventors: Ranjitha Gurunath Kulkarni, Ahmed Moustafa El Kholy, Ziad Al Bawab, Noha Alon, Imed Zitouni
  • Publication number: 20170295194
    Abstract: Systems and methods for evaluating the evaluation behaviors of an evaluator are presented. In contrast to evaluation methods that monitor and analyze click behaviors, the disclosed subject matter is directed to evaluating non-click behaviors. After obtaining results of an evaluation request submitted to a response service for evaluation by the evaluator, evaluation behaviors of the evaluator are monitored. The monitored evaluation behaviors are in association with an evaluation of the obtained results and one or more heuristics or rules are applied to the monitored evaluation behaviors to determining whether the monitored evaluation behaviors are within predetermined quality thresholds. If the monitored evaluation behaviors are not within the predetermined quality thresholds, the monitored evaluation behaviors are flagged as anomalous evaluation behaviors.
    Type: Application
    Filed: July 25, 2016
    Publication date: October 12, 2017
    Inventors: Imed Zitouni, Ahmed Awadallah, Bradley Paul Wethington, Aidan C Crook
  • Publication number: 20170293691
    Abstract: Examples of the present disclosure describe systems and methods of identifying good and bad abandonment using gesture movement. In aspects, user feedback signals may be received by a client device in response to the viewing and/or navigation of query results. The feedback signals may be provided to a framework for determining and/or analyzing query abandonment. The framework may identify gesture data in the feedback signals and extract feature data from the gesture data. The feature data may be provided to a metrics component to determine metrics and/or satisfaction values for the feature data. The metrics and/or feature data may be used to train a predictive model to identify good abandonment in query results.
    Type: Application
    Filed: August 8, 2016
    Publication date: October 12, 2017
    Applicant: Microsoft Technology Licensing, LLC
    Inventors: Imed Zitouni, Ahmed Hassan Awadallah, Aidan Crook, Bradley Wethington, Kyle Williams
  • Publication number: 20170060960
    Abstract: Examples of the present disclosure describe systems and methods for using online signals to improve judgment quality in Side-by-Side (SBS) evaluation. In aspects, two or more search result lists may be accessed within a query log. The search result lists may be used to generate and/or determine satisfaction metrics between the search result lists. The satisfaction metrics may be aggregated to automatically generate preference judgments for the search result lists. In some aspects, the preference judgments may be compared to the preference judgments of judges to measure the judgment quality of the judges.
    Type: Application
    Filed: August 28, 2015
    Publication date: March 2, 2017
    Applicant: MICROSOFT TECHNOLOGY LICENSING, LLC
    Inventors: Jin Kim, Imed Zitouni, Rajesh Patel
  • Publication number: 20170046431
    Abstract: Techniques for evaluating the quality of results obtained by a search engine. In an aspect, an evaluation platform utilizes task-level formulation to increase the accuracy of search result quality evaluation. Furthermore, initial queries may be reformulated until search results are deemed to satisfy the task description. Side-by-side comparison of results from multiple search engines is further provided to enhance the sensitivity of evaluation. Alternative aspects provide for collection of behavioral signals for training a classifier to classify the quality of an evaluator's feedback, as may be applied in, e.g., a crowd-sourcing context.
    Type: Application
    Filed: August 11, 2015
    Publication date: February 16, 2017
    Inventors: Imed Zitouni, Jinyoung Kim, Hyun Joon Jung, Ahmed Awadallah
  • Patent number: 9430533
    Abstract: Various technologies described herein pertain to evaluating search preferences. A search query, a first search result list returned by a first ranker system responsive to the search query, and a second search result list returned by a second ranker system responsive to the search query are received. A classifier is employed to predict (e.g., based upon values of features of the search query, the first search result list, and the second search result list) whether a search preference judgment (e.g., a side-by-side search preference judgment, etc.) indicates a quality difference between the first search result list and the second search result list. The search query, the first search result list, and the second search result list are excluded from a set of search queries and search result list pairs to be manually judged for search preference judgments when predicted to lack the quality difference.
    Type: Grant
    Filed: March 21, 2014
    Date of Patent: August 30, 2016
    Assignee: Microsoft Technology Licensing, LLC
    Inventors: Ahmed Hassan Awadallah, Imed Zitouni
  • Publication number: 20160147754
    Abstract: The claimed subject matter includes techniques for offline evaluation of ranking functions. An example system includes a first module configured to receive production log data, the first module to pre-process the production log data to generate an exploration data set. The example system also includes a second module configured to perform offline estimation of online metrics for ranking functions using the exploration data set. The example system also includes a third module to evaluate a proposed ranking function by comparing the estimated online metrics to a set of baseline metrics of a baseline ranking function and detecting that the estimated online metrics of the proposed ranking function exceed, are lower than, or are within a predetermined range of the baseline metrics.
    Type: Application
    Filed: November 21, 2014
    Publication date: May 26, 2016
    Inventors: Lihong Li, Jinyoung Kim, Imed Zitouni
  • Publication number: 20150379012
    Abstract: Described herein are technologies pertaining to determining which search engine results page (SERP), from a plurality of SERPs, is preferable to a user. A query is received, and multiple SERPS are retrieved based upon the query, wherein the multiple SERPs are generated independently from one another. Values of features of the query and the multiple SERPs are obtained, and a determination as to which of the SERPs from the multiple SERPs is preferable to the user is made based upon the values of the features. The SERP determined to be preferable over other SERPs in the multiple SERPs is presented to the user.
    Type: Application
    Filed: June 30, 2014
    Publication date: December 31, 2015
    Inventors: Ahmed Hassan Awadallah, Imed Zitouni, Frederic Behr
  • Publication number: 20150356489
    Abstract: Results, generated by human workers in response to HITs assigned to them, are evaluated based upon the behavior of the human workers in generating such results. Workers receive, together with an intelligence task to be performed, a behavior logger by which the worker's behavior is monitored while the worker performs the intelligence task. Machine learning is utilized to identify behavioral factors upon which the evaluation can be based and then to learn how to utilize such behavioral factors to evaluate the HIT results generated by workers, as well as the workers themselves. The identification of behavioral factors, and the subsequent utilization thereof, is informed by the behavior of, and corresponding results generated by, a trusted set of workers. Results evaluated to have been improperly generated can be discarded or simply downweighted. Workers evaluated to be operating improperly can be removed or retrained.
    Type: Application
    Filed: June 5, 2014
    Publication date: December 10, 2015
    Inventors: Gabriella Kazai, Imed Zitouni, Steven Shelford, Jinyoung Kim
  • Publication number: 20150269156
    Abstract: Various technologies described herein pertain to evaluating search preferences. A search query, a first search result list returned by a first ranker system responsive to the search query, and a second search result list returned by a second ranker system responsive to the search query are received. A classifier is employed to predict (e.g., based upon values of features of the search query, the first search result list, and the second search result list) whether a search preference judgment (e.g., a side-by-side search preference judgment, etc.) indicates a quality difference between the first search result list and the second search result list. The search query, the first search result list, and the second search result list are excluded from a set of search queries and search result list pairs to be manually judged for search preference judgments when predicted to lack the quality difference.
    Type: Application
    Filed: March 21, 2014
    Publication date: September 24, 2015
    Applicant: Microsoft Corporation
    Inventors: Ahmed Hassan Awadallah, Imed Zitouni
  • Publication number: 20150006199
    Abstract: Cascaded models may be applied to extract facts from a medical text. A first model may be applied to at least a portion of the medical text. The first model extracts at least one first medical fact. The at least one first medical fact is linked to at least first text in the at least a portion of the medical text. A second model may be applied to the first text. The second model extracts at least one second fact that is an attribute of the at least one first medical fact.
    Type: Application
    Filed: June 26, 2013
    Publication date: January 1, 2015
    Inventors: Neal E. Snider, Brian William Delaney, Girija Yegnanarayanan, Radu Florian, Martin Franz, Scott McCarley, John F. Pitrelli, Imed Zitouni, Salim E. Roukos
  • Patent number: 8903707
    Abstract: A method, an apparatus and an article of manufacture for determining a dropped pronoun from a source language. The method includes collecting parallel sentences from a source and a target language, creating at least one word alignment between the parallel sentences in the source and the target language, mapping at least one pronoun from the target language sentence onto the source language sentence, computing at least one feature from the mapping, wherein the at least one feature is extracted from both the source language and the at least one pronoun projected from the target language, and using the at least one feature to train a classifier to predict position and spelling of at least one pronoun in the target language when the at least one pronoun is dropped in the source language.
    Type: Grant
    Filed: January 12, 2012
    Date of Patent: December 2, 2014
    Assignee: International Business Machines Corporation
    Inventors: Bing Zhao, Imed Zitouni, Xiaoqiang Luo, Vittorio Castelli
  • Patent number: 8527262
    Abstract: Systems and methods are provided for automated semantic role labeling for languages having complex morphology. In one aspect, a method for processing natural language text includes receiving as input a natural language text sentence comprising a sequence of white-space delimited words including inflicted words that are formed of morphemes including a stem and one or more affixes, identifying a target verb as a stem of an inflicted word in the text sentence, grouping morphemes from one or more inflicted words with the same syntactic role into constituents, and predicting a semantic role of a constituent for the target verb.
    Type: Grant
    Filed: June 22, 2007
    Date of Patent: September 3, 2013
    Assignee: International Business Machines Corporation
    Inventors: Nandakishore Kambhatla, Imed Zitouni
  • Publication number: 20130185049
    Abstract: A method, an apparatus and an article of manufacture for determining a dropped pronoun from a source language. The method includes collecting parallel sentences from a source and a target language, creating at least one word alignment between the parallel sentences in the source and the target language, mapping at least one pronoun from the target language sentence onto the source language sentence, computing at least one feature from the mapping, wherein the at least one feature is extracted from both the source language and the at least one pronoun projected from the target language, and using the at least one feature to train a classifier to predict position and spelling of at least one pronoun in the target language when the at least one pronoun is dropped in the source language.
    Type: Application
    Filed: January 12, 2012
    Publication date: July 18, 2013
    Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION
    Inventors: Bing Zhao, Imed Zitouni, Xiaoqiang Luo, Vittorio Castelli
  • Patent number: 8412645
    Abstract: An exemplary processor-implemented method of determining whether a user of an online communication resource is an undesirable user includes the steps of building at least one model based on at least one feature of a feature set using at least one machine learning technique; and classifying the user by comparing at least one feature of the feature set that is associated with the user to the at least one model, a determination as to whether the user is an undesirable user being based at least in part on the classification of the user.
    Type: Grant
    Filed: May 30, 2008
    Date of Patent: April 2, 2013
    Assignee: International Business Machines Corporation
    Inventors: Ganesh N. Ramaswamy, Imed Zitouni