Search (6 results, page 1 of 1)

0.010750603 = product of:
  0.08600482 = sum of:
    0.08600482 = weight(_text_:unternehmen in 3547) [ClassicSimilarity], result of:
      0.08600482 = score(doc=3547,freq=4.0), product of:
        0.17271045 = queryWeight, product of:
          5.3116927 = idf(docFreq=592, maxDocs=44218)
          0.032515142 = queryNorm
        0.49797118 = fieldWeight in 3547, product of:
          2.0 = tf(freq=4.0), with freq of:
            4.0 = termFreq=4.0
          5.3116927 = idf(docFreq=592, maxDocs=44218)
          0.046875 = fieldNorm(doc=3547)
  0.125 = coord(1/8)

Abstract

Anwendungen von maschinellen Lernverfahren (ML) haben in jüngster Zeit aufsehenerregende Durchbrüche bei einer ganzen Reihe von Aufgaben in der maschinellen Sprachverarbeitung (NLP) erzielt. Der Fokus vieler Arbeiten liegt hierbei in der Entwicklung immer besserer Modelle, während der Anteil der Aufgaben in praktischen Projekten, der sich nicht mit Modellbildung, sondern mit Themen wie Datenbereitstellung sowie Evaluierung, Wartung und Deployment von Modellen beschäftigt, oftmals noch nicht ausreichend Beachtung erfährt. Im Ergebnis fehlen gerade Unternehmen, die nicht die Möglichkeit haben, eigene Plattformen für den Einsatz von ML und NLP zu entwerfen, oft geeignete Werkzeuge und Best Practices. Es ist zeichnet sich ab, dass in den kommenden Monaten eine gerade diesen praktischen Fragen zugewandte Ingenieurssicht auf ML und ihren Einsatz im Unternehmen an Bedeutung gewinnen wird.

0.0105228955 = product of:
  0.084183164 = sum of:
    0.084183164 = weight(_text_:union in 3042) [ClassicSimilarity], result of:
      0.084183164 = score(doc=3042,freq=4.0), product of:
        0.18718043 = queryWeight, product of:
          5.756716 = idf(docFreq=379, maxDocs=44218)
          0.032515142 = queryNorm
        0.44974342 = fieldWeight in 3042, product of:
          2.0 = tf(freq=4.0), with freq of:
            4.0 = termFreq=4.0
          5.756716 = idf(docFreq=379, maxDocs=44218)
          0.0390625 = fieldNorm(doc=3042)
  0.125 = coord(1/8)

Abstract

Based on State of the Union addresses from 1790 to 2014 (225 speeches delivered by 42 presidents), this paper describes and evaluates different text representation strategies. To determine the most important words of a given text, the term frequencies (tf) or the tf?idf weighting scheme can be applied. Recently, latent Dirichlet allocation (LDA) has been proposed to define the topics included in a corpus. As another strategy, this study proposes to apply a vocabulary specificity measure (Z?score) to determine the most significantly overused word-types or short sequences of them. Our experiments show that the simple term frequency measure is not able to discriminate between specific terms associated with a document or a set of texts. Using the tf idf or LDA approach, the selection requires some arbitrary decisions. Based on the term-specific measure (Z?score), the term selection has a clear theoretical basis. Moreover, the most significant sentences for each presidency can be determined. As another facet, we can visualize the dynamic evolution of usage of some terms associated with their specificity measures. Finally, this technique can be employed to define the most important lexical leaders introducing terms overused by the k following presidencies.

0.008868795 = product of:
  0.07095036 = sum of:
    0.07095036 = weight(_text_:unternehmen in 3977) [ClassicSimilarity], result of:
      0.07095036 = score(doc=3977,freq=2.0), product of:
        0.17271045 = queryWeight, product of:
          5.3116927 = idf(docFreq=592, maxDocs=44218)
          0.032515142 = queryNorm
        0.41080526 = fieldWeight in 3977, product of:
          1.4142135 = tf(freq=2.0), with freq of:
            2.0 = termFreq=2.0
          5.3116927 = idf(docFreq=592, maxDocs=44218)
          0.0546875 = fieldNorm(doc=3977)
  0.125 = coord(1/8)

Abstract

OpenAI hat einen Algorithmus zur Textgenerierung angeblich nicht vollständig veröffentlicht, weil er so gut sei und Missbrauch und Täuschung ermöglicht. Das u.a. von Elon Musk und Peter Thiel gegründete KI-Unternehmen OpenAI hatte im Februar erklärt, man habe den angeblich am weitesten fortgeschrittenen Algorithmus zur Sprachverarbeitung entwickelt. Der Algorithmus wurde lediglich anhand von 40 Gigabyte an Texten oder an 8 Millionen Webseiten trainiert, das nächste Wort in einem vorgegebenen Textausschnitt vorherzusagen. Damit könne man zusammenhängende, sinnvolle Texte erzeugen, die vielen Anforderungen genügen, zudem könne damit rudimentär Leseverständnis, Antworten auf Fragen, Zusammenfassungen und Übersetzungen erzeugt werden, ohne dies trainiert zu haben.

0.0016520072 = product of:
  0.013216058 = sum of:
    0.013216058 = product of:
      0.026432116 = sum of:
        0.026432116 = weight(_text_:22 in 1848) [ClassicSimilarity], result of:
          0.026432116 = score(doc=1848,freq=2.0), product of:
            0.113862485 = queryWeight, product of:
              3.5018296 = idf(docFreq=3622, maxDocs=44218)
              0.032515142 = queryNorm
            0.23214069 = fieldWeight in 1848, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              3.5018296 = idf(docFreq=3622, maxDocs=44218)
              0.046875 = fieldNorm(doc=1848)
      0.5 = coord(1/2)
  0.125 = coord(1/8)

Abstract

The goal of entity linking is to associate references to an entity that is found in unstructured natural language content to an authoritative inventory of known entities. This article describes the construction of 6 test collections for cross-language person-entity linking that together span 22 languages. Fully automated components were used together with 2 crowdsourced validation stages to affordably generate ground-truth annotations with an accuracy comparable to that of a completely manual process. The resulting test collections each contain between 642 (Arabic) and 2,361 (Romanian) person references in non-English texts for which the correct resolution in English Wikipedia is known, plus a similar number of references for which no correct resolution into English Wikipedia is believed to exist. Fully automated cross-language person-name linking experiments with 20 non-English languages yielded a resolution accuracy of between 0.84 (Serbian) and 0.98 (Romanian), which compares favorably with previously reported cross-language entity linking results for Spanish.

0.0013766728 = product of:
  0.011013382 = sum of:
    0.011013382 = product of:
      0.022026764 = sum of:
        0.022026764 = weight(_text_:22 in 1365) [ClassicSimilarity], result of:
          0.022026764 = score(doc=1365,freq=2.0), product of:
            0.113862485 = queryWeight, product of:
              3.5018296 = idf(docFreq=3622, maxDocs=44218)
              0.032515142 = queryNorm
            0.19345059 = fieldWeight in 1365, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              3.5018296 = idf(docFreq=3622, maxDocs=44218)
              0.0390625 = fieldNorm(doc=1365)
      0.5 = coord(1/2)
  0.125 = coord(1/8)

Date

2. 9.2014 21:22:48

0.0011013382 = product of:
  0.0088107055 = sum of:
    0.0088107055 = product of:
      0.017621411 = sum of:
        0.017621411 = weight(_text_:22 in 4217) [ClassicSimilarity], result of:
          0.017621411 = score(doc=4217,freq=2.0), product of:
            0.113862485 = queryWeight, product of:
              3.5018296 = idf(docFreq=3622, maxDocs=44218)
              0.032515142 = queryNorm
            0.15476047 = fieldWeight in 4217, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              3.5018296 = idf(docFreq=3622, maxDocs=44218)
              0.03125 = fieldNorm(doc=4217)
      0.5 = coord(1/2)
  0.125 = coord(1/8)

Date

22. 1.2018 11:32:44