Authorship attribution is the task of identifying the author of a given document.Various style markers have been proposed in the literature to deal with the authorship attribution task.Frequencies of function words have been shown to be very reliable and effective for this task.However, despite the fact that they are state-of-the-art, they basically rely on the invalid bag-of-words assumption, which stipulates that text is a set of independent words.In this contribution, we present a comparative study on using two different types of style marker based on function words for authorship attribution.We compare the effectiveness of using sequential rules of function words as style marker that do not relay on the bag-of-words assumption to that of the frequency of function words which does.Our results show that the frequencies of function words outperform the sequential rules.
In fine-grained sentiment analysis one has to deal with the composition of bi-polar phrases such as e.g. just punishment. Moreover, the top down prediction of phrase polarity as imposed by certain verbs on their direct objects sometimes is violated by the bottom up composed phrase polarity (e.g. ’to approve war’). We introduce a fine-grained polarity lexicon built along the lines of the Appraisal Theory and we investigate the composition of bi-polar phrases both, from a phrase internal point of view and from a verb-centered perspective. We have specified a multi-lingual polarity resource (French, English, German) and a system pipeline that carries out sentiment composition for these languages. We discuss examples with reference to each of these languages.
Data-driven parsers rely on recommendations from parse models, which are generated from a set of training data using a machine learning classifier, to perform parse operations. However, in some cases a parse model cannot recommend a parse action to a parser unless it learns from the training data what parse action(s) to take in every possible situation. Therefore, it will be hard for a parser to make an informed decision as to what parse operation to perform when a parse model recommends no/several parse actions to a parser. Here we examine the effect of various deterministic choices on a datadriven parser when it is presented with no/several recommendation from a parse model.
Question answering systems generally, and Arabic systems are no exception, hit an upper bound of performance due to the propagation of error in their pipeline. This increases the significance of answer selection systems as they enhance the certainty and accuracy of question answering. Very few works tackled the Arabic answer selection problem, and they did not demonstrate encouraging performance because they use the same question answering pipeline without any changes to satisfy the requirements of answer selection. In this paper, we present “ALQASIM 2.0”, which uses a new approach to Arabic answer selection. It analyzes the reading test documents instead of the questions, utilizes sentence splitting, root expansion, and semantic expansion using an automatically generated ontology. Our experiments are conducted on the test-set provided by CLEF 2012 through the task of QA4MRE. This approach leads to a promising performance of 0.36 accuracy and 0.42 c@1.
You Shall Find the Target Via its Companion Words: Specifications of a Navigational Tool to Help Authors to Overcome the Tip-Of-The-Tongue Problem was published in Natural Language Processing and Cognitive Science on page 215.
We recently equipped the open-source spoken dialog system (SDS) Halef with the speech synthesizer Festival which supports both unit selection and HMM-based voices. Inspired by the most recent Blizzard Challenge, the largest international speech synthesis competition, we sought to find which of the freely available voices in Festival and those of the strongest competitor Mary are promising candidates for operational use in Halef. After conducting a subjective evaluation involving 36 participants, we found that Festival was clearly outperformed by Mary and that unit selection voices performed en par, if not better, than HMM-based ones.
