Dataset Viewer
Auto-converted to Parquet Duplicate
speaker_id
stringclasses
12 values
english_audio_path
audioduration (s)
3.99
17.8
english_transcript
stringlengths
30
266
arabic_cloned_path
audioduration (s)
1.92
60.5
arabic_transcript
stringclasses
49 values
2023.acl-long.14
In the framework of construction grammar, grammatical constructions are syntactic forms that are
حوسبة اللغة: ما بين هيمنة الشبكات العصبية وتراجع نماذج ماركوف المخفية
2023.acl-long.14
conventionally paired with semantic and pragmatic meanings. Individual words fit the criteria,
منذ اكتشافها في بدايات القرن الماضي، حظيت نماذج (ماركوف) على قبول واستحسان الباحثين، واستخدمت في مجالات متعددة.
2023.acl-long.14
for constructions, but so do units larger than words, such as fixed idioms and templatic expressions.
وتشمل نماذج ماركوف نوعين رئيسين: نماذج ماركوف الظاهرة، ونماذج ماركوف المخفية.
2023.acl-long.14
An example of a templatic construction is the adjective-adjective-noun-noun construction. For example, the family spent
تم الاستعانة بقواعد البيانات الرقمية لجمع أكبر عدد ممكن من الأوراق العلمية، ثم فُرزت لاختيار قائمة قصيرة من الأوراق المتعلقة بحوسبة اللغات، ثم صُنِّفت حسب فترة النشر.
2023.acl-long.14
a beautiful three days in London, or a lucky three students will get scholarships. This construction has been
أثر هندسة التلقين لأدوات الذكاء الاصطناعي التوليدي في جودة إنتاج المحتوى النصي التسويقي العربي
2023.acl-long.14
been argued to construe its noun phrase arguments as a single atomic unit of measure.
هدفت هذه الدراسة إلى تقييم أثر هندسة التلقين لأدوات الذكاء الاصطناعي التوليدي في إنتاج المحتوى النصي التسويقي العربي من وجهة نظر طلاب التسويق في جامعة القدس المفتوحة في فلسطين، وذلك بالاعتماد على المنهج الوصفي التحليلي، باستخدام استبانة إلكترونية وُزِّعت على عينة عشوائية بسيطة تتكون من (92) مشاركاً.
2023.acl-long.14
Construction grammarians such as Adele Goldberg argue that even very basic syntactic phenomena like grammatical subjects
وباستخدام SPSS، في ضوء ما توصلت إليه هذه الدراسة من نتائج، يمكن التأكيد على أن هندسة التلقين لأدوات الذكاء الاصطناعي التوليدي تمثل مدخلاً أساسياً ومؤثراً في جودة إنتاج المحتوى النصي التسويقي العربي.
2023.acl-long.14
and objects constitute a grammatical subject hood and object hood constitute constructions that is they constitute
حيث أظهرت التحليلات الإحصائية أن الأبعاد الخمسة لهندسة التلقين — وهي: صياغة المهام، كتابة السياق، إعطاء الأدوار، مؤشر المخرجات، وتزويد المدخلات — تلعب جميعها أدواراً مهمة ومتكاملة في تحسين فاعلية النصوص التسويقية المنتجة.
2023.acl-long.14
to form meaning pairings. So how can we get at the meaning of grammatical constructions through distributional
وقد تبيّن أن أعلى الأبعاد تأثيراً كان "مؤشر المخرجات" بنسبة (52%)، يليه "كتابة السياق" بنسبة (47.9%)، ثم "تزويد المدخلات" بنسبة (46.0%)، ثم "إعطاء الأدوار" بنسبة (38.1%).
2023.acl-long.14
modeling? If subjecthood itself is a construction, what meaning is it paired with? Our paper at ACL
وتؤكد هذه النتائج ضرورة نقل المفهوم من مجرد "استخدام أداة" إلى إدارة تفاعلية واعية مع الأداة عبر هندسة تلقين فعّالة.
2023.acl-long.14
a method for approaching this question and applies it to two different grammatical constructions. We frame the problem in terms of
كما أوصت الدراسة بدمج مفاهيم هندسة التلقين في المناهج الجامعية ضمن مساقات التسويق الرقمي وتصميم برامج تدريبية تطبيقية على الأبعاد الخمسة لهندسة التلقين، وتشجيع البحوث التجريبية التي تختبر مدى فاعلية المحتوى الناتج من الذكاء الاصطناعي في حملات تسويقية حقيقية.
2023.acl-long.14
construal? Can we use language models to discover how constructions construe their arguments in particular?
معالجة النصوص العربية لغايات تطبيقات المعالجة الآلية اللغات الطبيعية
2023.acl-long.14
ways. This paper represents joint work with my advisor Katrina Erck and co-author Kyle Mahold at the University of Texas
يهدف هذا البحث إلى وصف مقاربة جديدة لمعالجة النصوص العربية المشكلة وغير المشكلة من أجل تهيئتها للاستعمال في تطبيقات المعالجة الآلية للغات الطبيعية.
2023.acl-long.14
at Austin. What we do is we take uninterpretable contextual word embeddings from pre-trained language models
بُنيت المقاربة الجديدة على قواعد محددة مسبقاً وتتكون من أربعة مراحل: فصل المكون الأساسي في النص (الكلمة)، فصل جذور الكلمات، التحليل الصرفي للكلمات، وإضافة توصيف للكلمات على النص.
2023.acl-long.14
our case BERT embeddings, and we train mappings from that embedding space into an interpreter
تعالج المرحلة الأولى النص بهدف عزل الكلمات وإعادة تمثيلها بطريقة معيارية.
2023.acl-long.14
embedding space where each dimension of the space corresponds to a different semantic feature like small
تشمل المرحلة الثالثة محللاً صرفياً مبنياً على قواعد محددة، يستخرج جذر والنمط الصرفي لكل كلمة.
2023.acl-long.14
yellow, is a fruit, grows on trees, etc.
أما المرحلة الأخيرة فتضيف توصيفات على النص تشمل الخصائص الصرفية لكل كلمة.
2023.acl-long.14
and these data come from semantic feature norms. Once trained on known feature norms at the word type,
تستطيع الطريقة المقترحة في هذا البحث التعامل مع النصوص العربية المشكلة وغير المشكلة وتنتج لكل كلمة من النص مجموعة من المعلومات اللغوية الضرورية لكثير من التطبيقات.
2023.acl-long.14
level, our models can predict semantic features for a single word in context. For example,
لقد تم تصميم المعالج بحيث يمكن استخدامه مع الكثير من تطبيقات المعالجة الآلية للغات الطبيعية مثل ترجمة النصوص، تلخيص النصوص، تصحيح النصوص، استخراج المعلومات والتشكيل الآلي للنصوص العربية.
2023.acl-long.14
the model predicts different features for a ring that one would wear on one's finger and a ring that one fights in, which it
كفاءة الذكاء الاصطناعي في الإجابة عن أسئلة الثانوية العامة (توجيهي فلسطيني) في المقررات العربية: دراسة حالة منصة شات.جي.بي.تي ومنصة جيميني
2023.acl-long.14
gets higher values for features like sport and place. Others have begun to study the use of
التعرف على كفاءة الذكاء الاصطناعي في الإجابة عن أسئلة التوجيهي في المقررات العربية والمقارنة بين أداء النماذج اللغوية الأكثر شهرة لهذا الغرض.
2023.acl-long.14
feature norms to study contextual semantic space. Our paper extends the task of contextual
تم استخدام المنهج الوصفي التحليلي بالأسلوب النوعي والأسلوب الكمي، واستخدمت عينة قصدية من امتحانات الثانوية العامة (التوجيهي) للأعوام (2021، 2022، 2023) وعينة من عشرة معلمين من التوجيهي، كما تم استخدام كشف تصحيح امتحانات التربية الإسلامية واللغة العربية لتقييم إجابات الذكاء الاصطناعي، كما تم استخدام الرزمة الإحصائية (SP...
2023.acl-long.14
feature prediction to new feature norm data sets, offers a rigorous testing of different mapping methods for different data sets,
أن النماذج اللغوية في منصة شات.جي.بي.تي (ChatGPT) ومنصة جيميني (Gemini) استطاعا الإجابة على أسئلة التوجيهي التي طرحت عليها ضمن مقرر اللغة العربية ومقرر التربية الإسلامية، وبدرجات مختلفة تتراوح ما بين المتوسطة والجيدة، إلا أن منصة جيميني كانت الأكثر كفاءة وقدرة على الإجابة عن الأسئلة المطروحة عليها بشكل أكثر صحة وبدرجة ...
2023.acl-long.14
and devises two new methods for evaluating contextual feature prediction. But what we think is really neat is that we
أوصت الدراسة بالعمل على تحسين قاعدة بيانات التدريب للنماذج اللغوية بإضافة مزيد من النصوص والمواد التعليمية المحددة المرتبطة بالمناهج الدراسية في فلسطين.
2023.acl-long.14
can then use this method to study grammatical constructions. The first construction we look at is the AANN construction.
منذ إنشاء أول كمبيوتر، واجه البشر صعوبة في التفاعل مع الآلات، والمعالجة اللغوية الطبيعية (NLP) هي الوسيلة التي يتحدث من خلالها البشر العصريون مع الآلات، بطريقة يفهمونها، بتطبيق مفاهيم الذكاء الاصطناعي، لتحويل اللغة البشرية المعقدة، إلى أرقام تفهمها الآلات.
2023.acl-long.14
To look at this construction, we use templates to generate two sets of sentences, one using the construction,
وتعد اللغة العربية أحد ميادين البرمجة اللغوية الوطنية، التي لم تزدهر بعد بشكل كامل وتصبح فعالة بما فيه الكفاية لمقارنتها مع اللغات الأخرى مثل اللغة الإنجليزية.
2023.acl-long.14
a surprising 20 soldiers greeted us at the door, and one using the canonical word order,
ولهذا السبب تركز هذه الورقة على ANLP، في محاولة للبحث في هذا المجال، باستخدام مجموعة بيانات أصلية باللغة العربية فقط، لإنشاء وتدريب نماذج مختلفة لتحليل النصوص.
2023.acl-long.14
forming a near minimal pair. 20 surprising soldiers greeted us at the door. Then we run these sentences
ويبين التحقيق الذي أجري على ثلاثة أنواع من تحليل النصوص، وهي: الكشف العاجل، وتحليل المشاعر، وتصنيف المواضيع، أن الكشف عن الإلحاح هو أبسط ثلاثة نماذج، وهو النموذج الأفضل، بدقة 87 في المائة، فيما يتعلق بتحليل المشاعر، تظهر النتائج أن اكتشاف المشاعر باللغة العربية هو اكتشاف صعب، حيث أن نموذج "التوجه" لديه دقة نهائية تبلغ ...
2023.acl-long.14
through the model and look at the average feature vector that comes out for the sort of average slot filler
وأخيرا، يبدو أن تصنيف الموضوعات باللغة العربية يمثل مشكلة تتعلق بحجم مجموعة البيانات وتعقيدها، حيث أن مجموعة البيانات تزداد حجما وتصبح أقل تعقيدا، وتزداد الدقة بسرعة، وتزداد دقة المواضيع مع مجموعات البيانات الأكبر حجما بشكل ملحوظ، وتبين أيضا أن دقة المواضيع الأقل تعقيدا، التي تحتوي على مجموعة أقل تنوعا من المواضيع الفر...
2023.acl-long.14
for the noun in this construction. So what are the average semantic features for the head noun? And then we do the same thing for the noun
تخيّل إمكانية إنشاء روبوت دردشة قادر على التفاعل مع المستخدمين وأتمتة المهام والتعامل مع العمليات المعقدة—كل ذلك دون كتابة أي سطر برمجي.
2023.acl-long.14
in the canonical unmarked word order? What are the average features for the noun? And then we compare the delta
FlowX يتيح ذلك من خلال منصة مبتكرة لبناء روبوتات دردشة بطريقة مرئية، تمكّن المستخدمين، بغض النظر عن خبرتهم التقنية، من تصميم ونشر روبوتات دردشة مخصّصة لمجالات متعددة، بدءًا من خدمة العملاء البسيطة وحتى مهام الأتمتة المتقدمة.
2023.acl-long.14
between the aaa and n features and the canonical features. When we look at differences between the aaaa and n nouns and the
توفر المنصة أيضًا سوقًا مدمجًا يتيح للمستخدمين مشاركة واستكشاف الروبوتات، كما تمنح المطورين القدرة على تخصيص المظهر البصري للروبوتات التي يتفاعل معها المستخدمون النهائيون.
2023.acl-long.14
same noun in the default word ordering, we see a pattern to the features with the highest shift of the top five
ما يميز FlowX هو بنيتها المتقدمة القائمة على الرسوم البيانية، والتي تتيح للمستخدمين بناء تدفقات محادثة معقدة باستخدام أنواع خاصة من الوحدات عبر واجهة سحب وإفلات.
2023.acl-long.14
features that are most salient to AANN we see features measure one and unit so
يمكن للمستخدمين إنشاء سير عمل باستخدام وحدات لمعالجة النصوص، ودمج نماذج اللغة الكبيرة (LLMs)، بالإضافة إلى وحدات تنفيذ طلبات API لأتمتة مهام لا محدودة.
2023.acl-long.14
In this construction, nouns like days are higher and measurement unit features
النتيجة هي شجرة قرارات ((Decision Tree متقدمة كان من الصعب جدًا إدارتها برمجيًا يدويًا.
2023.acl-long.14
And we think this tells us something about how the word days or even the entire noun phrase is construed in this sentence.
تعتمد FlowX على بنية (Microservices)، وتشمل خدمة API مبنية بـASP.NET Core لإدارة سير العمل، ومنفذ LangChain بلغة بايثون لتنفيذ عمليات LLM، وواجهة أمامية مبنية بـReact Native لتوفير تجربة تطوير سهلة.
2023.acl-long.14
as a unit of measure. This finding accords with the semantics literature on the AANN construction,
تم تصميم نموذج نطاق (Domain Model) يدعم إدارة المستخدمين ووظائف الروبوت، ويستخدم هيكل رسومي يميّز بين روابط التدفق (للتحكم في تسلسل المحادثة) وروابط البيانات (لنقل المعلومات بين العقد).
2023.acl-long.14
where it's observed that AANN arguments must be interpreted as indivisible atomic units.
يدعم النظام إدارة حالة المحادثة ويوفر تجربة تطوير متكاملة من إنشاء سير العمل، إلى التخصيص المرئي، إلى الاختبار المباشر.
2023.acl-long.14
The other construction we look at is grammatical subject and objecthood. In this case, we construct near minimal pairs by taking natural
كما تم تحسين الأداء لضمان عرض فعال لسير العمل المعقد واستجابة سريعة في تفاعلات الروبوت.
2023.acl-long.14
occurring English sentences and then swapping the subject and object. So we use the same method
نعتمد أيضًا على تقنيات مثل LangChain لتنفيذ المهام وتجريد التعقيدات التقنية، مما يقلل الحاجة إلى الصيانة من جانب المستخدم.
2023.acl-long.14
and look at natural subjects and see that they have higher
لاحظنا أن الحلول الحالية تنقسم إلى فئتين رئيسيتين: أدوات البناء المرئية مثل Chatfuel وManyChat التي تركز على التدفقات المنظمة لكنها تفتقر إلى قدرات الذكاء الاصطناعي المتقدمة، ومنصات الذكاء الاصطناعي مثل Dialogflow التي تتطلب خبرة تقنية.
2023.acl-long.14
feature, higher values for features associated with animacy, like biomotion, body, human,
FlowX يملأ هذه الفجوة عبر توفير أداة بناء مرئية متاحة تدمج بين النهجين بسلاسة.
2023.acl-long.14
and face and objects have lower values for these features. And then when you switch place,
بخلاف المنافسين، يقدم FlowX روبوتات هجينة تجمع بين القدرة على التحكم في المحادثة عند الحاجة والاستفادة من نماذج اللغة للتعامل مع المدخلات غير المتوقعة، وذلك ضمن بيئة تطوير موحّدة.
2023.acl-long.14
same word when moved from subject to object position will decrease in its animacy features.
باختصار، FlowX يعيد تعريف تطوير روبوتات الدردشة من خلال نهجه الهجين المبتكر.
2023.acl-long.14
There are several advantages for this method. The first is that we can use the same model, which is available online,
فهو يمكّن المستخدمين من تصميم سير عمل مرئي يجمع بين المنطق المنظم والذكاء الاصطناعي الديناميكي، والتكامل مع الخدمات الخارجية عبر طلبات API لأتمتة المهام، وحتى تخصيص مظهر الروبوت، مما يتيح إنشاء روبوتات دردشة من الجيل القادم تتميز بالقوة وسهولة الاستخدام.
2023.acl-long.14
study many different phenomena of linguistic interest, and we can avoid topical confounds by using
تحتوي اللغة العربية على عدد محدد من مجموعات الحروف (particles)، مثل: حروف الجر، وحروف الجزم، وحروف النصب، الخ. وتؤدي كل مجموعة دوراً معيناً في السياق الذي ترد فيه.
2023.acl-long.14
near-minimal pair sentence data sets, and there's no need to select ahead of time what
بشكل عام، فإن الوسم هو عملية تحديد الصيغة المناسبة لكل كلمة سواء أكانت اسماً أو فعلاً أو حرفاً، وذلك بالاعتماد على السياق الذي ترد فيه.
2023.acl-long.14
features we're interested in or to train a bespoke classifier for each feature of interest. We could just look at which ones.
وكذلك يستخدم لتحليل النص لفهم المعنى المطلوب لاستخدامه في محركات البحث (search engines).
2023.acl-long.14
are most important to each difference that we're interested in.
تشمل تطبيقات معالجة اللغات أيضاً الترجمة الآلية (machine translation)، تأليف الكلام (speech synthesis)، التعرف على الكلام (speech recognition)، التشكيل الآلي (diacritization)، وغيرها.
2023.acl-long.809
Hello! Welcome to my presentation. I'm going to talk about script normalization for our conventional writing of under-resourced languages, particularly in multilingual or bilingual communities. This work was carried out at George Mason University with Adonis and
حوسبة اللغة: ما بين هيمنة الشبكات العصبية وتراجع نماذج ماركوف المخفية
2023.acl-long.809
and will be presented at ACL 2023 in Toronto, Canada. Let me start by providing some facts to just define what unconventional writing is. As you know, there are more than 7,000 languages
منذ اكتشافها في بدايات القرن الماضي، حظيت نماذج (ماركوف) على قبول واستحسان الباحثين، واستخدمت في مجالات متعددة.
2023.acl-long.809
spoken around the world. And there are almost 300 writing systems that are in use as of today. Interestingly, less than 4,000 languages have a written form. This means that a lot of languages around the world have
وتشمل نماذج ماركوف نوعين رئيسين: نماذج ماركوف الظاهرة، ونماذج ماركوف المخفية.
2023.acl-long.809
longer oral tradition than a written one here on the map you can see some of the most widespread language script families like Arabic one or Cyrillic or Latin script so here's what I'm thinking
تم الاستعانة بقواعد البيانات الرقمية لجمع أكبر عدد ممكن من الأوراق العلمية، ثم فُرزت لاختيار قائمة قصيرة من الأوراق المتعلقة بحوسبة اللغات، ثم صُنِّفت حسب فترة النشر.
2023.acl-long.809
thinking about what happens to the speakers of languages that are spoken in multilingual or bilingual communities. You can think of community as a country like Pakistan. What happens to all these languages that are
أثر هندسة التلقين لأدوات الذكاء الاصطناعي التوليدي في جودة إنتاج المحتوى النصي التسويقي العربي
2023.acl-long.809
spoken here and only Urdu and English are official languages. Or, in the case of India, even though many of these languages are officially recognized, how people write in these countries. Or in the case of Iraq, where Kurdish
هدفت هذه الدراسة إلى تقييم أثر هندسة التلقين لأدوات الذكاء الاصطناعي التوليدي في إنتاج المحتوى النصي التسويقي العربي من وجهة نظر طلاب التسويق في جامعة القدس المفتوحة في فلسطين، وذلك بالاعتماد على المنهج الوصفي التحليلي، باستخدام استبانة إلكترونية وُزِّعت على عينة عشوائية بسيطة تتكون من (92) مشاركاً.
2023.acl-long.809
and Arabic are the two official ones even though there are many other languages spoken or in Iran which is a fascinating case in terms of language diversity yet for some weird reason only Persian is recognized as an official language so
وباستخدام SPSS، في ضوء ما توصلت إليه هذه الدراسة من نتائج، يمكن التأكيد على أن هندسة التلقين لأدوات الذكاء الاصطناعي التوليدي تمثل مدخلاً أساسياً ومؤثراً في جودة إنتاج المحتوى النصي التسويقي العربي.
2023.acl-long.809
To define what happens within these communities, these speakers, that the majority of them have not received education in their mother language, we define something which is called unconventional writing. And unconventional writing refers to the
حيث أظهرت التحليلات الإحصائية أن الأبعاد الخمسة لهندسة التلقين — وهي: صياغة المهام، كتابة السياق، إعطاء الأدوار، مؤشر المخرجات، وتزويد المدخلات — تلعب جميعها أدواراً مهمة ومتكاملة في تحسين فاعلية النصوص التسويقية المنتجة.
2023.acl-long.809
phenomenon where the speaker of a language, instead of using the script or orthography of their own language, write in the script orthography of the dominant language of the country or the community. Let's have an example. Someone who is
وقد تبيّن أن أعلى الأبعاد تأثيراً كان "مؤشر المخرجات" بنسبة (52%)، يليه "كتابة السياق" بنسبة (47.9%)، ثم "تزويد المدخلات" بنسبة (46.0%)، ثم "إعطاء الأدوار" بنسبة (38.1%).
2023.acl-long.809
a native speaker of Punjabi, instead of writing in their own script of Punjabi, write in the Urdu script. And here, we have a lot of problems that are underestimated and usually not discussed much in NLP. They have not been discussed.
وتؤكد هذه النتائج ضرورة نقل المفهوم من مجرد "استخدام أداة" إلى إدارة تفاعلية واعية مع الأداة عبر هندسة تلقين فعّالة.
2023.acl-long.809
discussed much in NLP, even though they are open and serious problems. One of the problems is due to unsystematic writing. We don't know if the writer tries to write according to orthography of the donor,
كما أوصت الدراسة بدمج مفاهيم هندسة التلقين في المناهج الجامعية ضمن مساقات التسويق الرقمي وتصميم برامج تدريبية تطبيقية على الأبعاد الخمسة لهندسة التلقين، وتشجيع البحوث التجريبية التي تختبر مدى فاعلية المحتوى الناتج من الذكاء الاصطناعي في حملات تسويقية حقيقية.
2023.acl-long.809
script or the script of their own language because every speaker can come up with a way of writing right it's not it's not systematic to be able to map it or apply some techniques that are known in
معالجة النصوص العربية لغايات تطبيقات المعالجة الآلية اللغات الطبيعية
2023.acl-long.809
transliteration, basically mapping from one script to another one. In the context of code switching, we are also dealing with words that are written as they are in the donor language and without being rewritten in the way that
يهدف هذا البحث إلى وصف مقاربة جديدة لمعالجة النصوص العربية المشكلة وغير المشكلة من أجل تهيئتها للاستعمال في تطبيقات المعالجة الآلية للغات الطبيعية.
2023.acl-long.809
the speaker says or pronounces it so there's no specific rules to be able to deal with unconventional writing here are some examples for example writing Arabic in the in the Latin script this is this has been around for
بُنيت المقاربة الجديدة على قواعد محددة مسبقاً وتتكون من أربعة مراحل: فصل المكون الأساسي في النص (الكلمة)، فصل جذور الكلمات، التحليل الصرفي للكلمات، وإضافة توصيف للكلمات على النص.
2023.acl-long.809
for some time right now, I think more than a decade. It's called Arabizi or writing Greek in English or writing French as if it is an SMS where we try to abbreviate stuff and come up with conventions.
تعالج المرحلة الأولى النص بهدف عزل الكلمات وإعادة تمثيلها بطريقة معيارية.
2023.acl-long.809
These are all types of some examples of unconventional writing. So in this study, I am particularly interested in the perso-rabic scripts. So basically the light green family, like the countries that are present.
تشمل المرحلة الثالثة محللاً صرفياً مبنياً على قواعد محددة، يستخرج جذر والنمط الصرفي لكل كلمة.
2023.acl-long.809
specified in light green. Arabic itself, or the Arabic script, we should say, has been traditionally used for writing Arabic, even right now for modern Arabic. It has been around for over a millennium. It has been commonly used and adopted by a lot of other language
أما المرحلة الأخيرة فتضيف توصيفات على النص تشمل الخصائص الصرفية لكل كلمة.
2023.acl-long.809
communities, including Persian and Urdu. So these two languages have been around for, have been written for quite a few centuries, and they have adopted the basic or the
تستطيع الطريقة المقترحة في هذا البحث التعامل مع النصوص العربية المشكلة وغير المشكلة وتنتج لكل كلمة من النص مجموعة من المعلومات اللغوية الضرورية لكثير من التطبيقات.
2023.acl-long.809
original modern Arabic script to represent some of the phonemes that exist in the languages. And this has been done for some other languages in Iran or Iraq, like Guarani, Kurdish,
لقد تم تصميم المعالج بحيث يمكن استخدامه مع الكثير من تطبيقات المعالجة الآلية للغات الطبيعية مثل ترجمة النصوص، تلخيص النصوص، تصحيح النصوص، استخراج المعلومات والتشكيل الآلي للنصوص العربية.
2023.acl-long.809
And languages in Pakistan and India are like Sindhi, Kashmiri, Punjabi, let alone some other languages that have traditionally relied on the Arabic script but have changed recently.
كفاءة الذكاء الاصطناعي في الإجابة عن أسئلة الثانوية العامة (توجيهي فلسطيني) في المقررات العربية: دراسة حالة منصة شات.جي.بي.تي ومنصة جيميني
2023.acl-long.809
So these are the type of languages that I am interested about because they are spoken in multilingual communities and the phenomenon of unconventional writing is more widespread. This is a little bit like the Pandora's box.
التعرف على كفاءة الذكاء الاصطناعي في الإجابة عن أسئلة التوجيهي في المقررات العربية والمقارنة بين أداء النماذج اللغوية الأكثر شهرة لهذا الغرض.
2023.acl-long.809
which is the source of all problems if we deal with this type of data in NLP applications. So to tackle this problem, this is what we define. We define script normalization. Script normalization refers to the task
تم استخدام المنهج الوصفي التحليلي بالأسلوب النوعي والأسلوب الكمي، واستخدمت عينة قصدية من امتحانات الثانوية العامة (التوجيهي) للأعوام (2021، 2022، 2023) وعينة من عشرة معلمين من التوجيهي، كما تم استخدام كشف تصحيح امتحانات التربية الإسلامية واللغة العربية لتقييم إجابات الذكاء الاصطناعي، كما تم استخدام الرزمة الإحصائية (SP...
2023.acl-long.809
of normalizing of a text written in an unconventional script as i described by trying to map everything to the conventional script or orthography of the language the task has a lot in common with
أن النماذج اللغوية في منصة شات.جي.بي.تي (ChatGPT) ومنصة جيميني (Gemini) استطاعا الإجابة على أسئلة التوجيهي التي طرحت عليها ضمن مقرر اللغة العربية ومقرر التربية الإسلامية، وبدرجات مختلفة تتراوح ما بين المتوسطة والجيدة، إلا أن منصة جيميني كانت الأكثر كفاءة وقدرة على الإجابة عن الأسئلة المطروحة عليها بشكل أكثر صحة وبدرجة ...
2023.acl-long.809
transliteration, spelling error correction, and text normalization. The three of these tasks are somehow described and well defined in the literature, but script normalization is not something that we notice in the literature
أوصت الدراسة بالعمل على تحسين قاعدة بيانات التدريب للنماذج اللغوية بإضافة مزيد من النصوص والمواد التعليمية المحددة المرتبطة بالمناهج الدراسية في فلسطين.
2023.acl-long.809
That's why we are defining it as it is in our paper. To tackle this problem, this is our approach. We first collect data for the selected languages. It has not been easy because some of them are not commonly written in the Perso-Arabic script.
منذ إنشاء أول كمبيوتر، واجه البشر صعوبة في التفاعل مع الآلات، والمعالجة اللغوية الطبيعية (NLP) هي الوسيلة التي يتحدث من خلالها البشر العصريون مع الآلات، بطريقة يفهمونها، بتطبيق مفاهيم الذكاء الاصطناعي، لتحويل اللغة البشرية المعقدة، إلى أرقام تفهمها الآلات.
2023.acl-long.809
like Kurmanji Kurdish which uses the Latin based script and some of them don't have much data available online so we had to collect our data elsewhere other than the Wikipedia pages of the languages
وتعد اللغة العربية أحد ميادين البرمجة اللغوية الوطنية، التي لم تزدهر بعد بشكل كامل وتصبح فعالة بما فيه الكفاية لمقارنتها مع اللغات الأخرى مثل اللغة الإنجليزية.
2023.acl-long.809
Then we try to synthetically generate noisy data because we don't have access to genuinely unconventionally written because of lack of technology, obviously. So what we do is that we synthetically generate data
ولهذا السبب تركز هذه الورقة على ANLP، في محاولة للبحث في هذا المجال، باستخدام مجموعة بيانات أصلية باللغة العربية فقط، لإنشاء وتدريب نماذج مختلفة لتحليل النصوص.
2023.acl-long.809
to later train our models for these tasks. And to do so, we have two steps. One, we try to map the script of one language to the other one, somehow trying to mime
ويبين التحقيق الذي أجري على ثلاثة أنواع من تحليل النصوص، وهي: الكشف العاجل، وتحليل المشاعر، وتصنيف المواضيع، أن الكشف عن الإلحاح هو أبسط ثلاثة نماذج، وهو النموذج الأفضل، بدقة 87 في المائة، فيما يتعلق بتحليل المشاعر، تظهر النتائج أن اكتشاف المشاعر باللغة العربية هو اكتشاف صعب، حيث أن نموذج "التوجه" لديه دقة نهائية تبلغ ...
2023.acl-long.809
how a speaker would write um unconventionally and by for that we try to find what are the common characters what are the visually similar characters in the two scripts and
وأخيرا، يبدو أن تصنيف الموضوعات باللغة العربية يمثل مشكلة تتعلق بحجم مجموعة البيانات وتعقيدها، حيث أن مجموعة البيانات تزداد حجما وتصبح أقل تعقيدا، وتزداد الدقة بسرعة، وتزداد دقة المواضيع مع مجموعات البيانات الأكبر حجما بشكل ملحوظ، وتبين أيضا أن دقة المواضيع الأقل تعقيدا، التي تحتوي على مجموعة أقل تنوعا من المواضيع الفر...
2023.acl-long.809
if there are any orthographic rules that can be applied and doing so we create a map that tries to map every character in the source language in this case for example in Kurdish to Arabic or in Kashmiri to Urdu
تخيّل إمكانية إنشاء روبوت دردشة قادر على التفاعل مع المستخدمين وأتمتة المهام والتعامل مع العمليات المعقدة—كل ذلك دون كتابة أي سطر برمجي.
2023.acl-long.809
We also include some data from dictionaries. We look at data in dictionaries and see how common or more probable our mappings are. And if something is 100% sure to be mapped to another character,
FlowX يتيح ذلك من خلال منصة مبتكرة لبناء روبوتات دردشة بطريقة مرئية، تمكّن المستخدمين، بغض النظر عن خبرتهم التقنية، من تصميم ونشر روبوتات دردشة مخصّصة لمجالات متعددة، بدءًا من خدمة العملاء البسيطة وحتى مهام الأتمتة المتقدمة.
2023.acl-long.809
we set a probability of 1 to it. So using this script mapping and character alignment matrix, we then synthetically generate data, meaning that we get a clean data in our data set for, let's say, Kurdish in the case of this example.
توفر المنصة أيضًا سوقًا مدمجًا يتيح للمستخدمين مشاركة واستكشاف الروبوتات، كما تمنح المطورين القدرة على تخصيص المظهر البصري للروبوتات التي يتفاعل معها المستخدمون النهائيون.
2023.acl-long.809
And we try to inject noise synthetically by a certain level of noise from 20% to 100%. And this is a simple replacement and substitution of characters by noisy ones.
ما يميز FlowX هو بنيتها المتقدمة القائمة على الرسوم البيانية، والتي تتيح للمستخدمين بناء تدفقات محادثة معقدة باستخدام أنواع خاصة من الوحدات عبر واجهة سحب وإفلات.
2023.acl-long.809
as if the clean text is written in an unconventional way. Finally, we use our data under different setups to train a transformer model, an encoder-decoder model, where given a noisy or unconventionally written
يمكن للمستخدمين إنشاء سير عمل باستخدام وحدات لمعالجة النصوص، ودمج نماذج اللغة الكبيرة (LLMs)، بالإضافة إلى وحدات تنفيذ طلبات API لأتمتة مهام لا محدودة.
2023.acl-long.809
text the model tries to learn to normalize the given input so this is our approach and these are our findings we basically notice that our model our normalization models can to a very good extent
النتيجة هي شجرة قرارات ((Decision Tree متقدمة كان من الصعب جدًا إدارتها برمجيًا يدويًا.
2023.acl-long.809
extent prevent or tackle the problem of unconventional writing by normalizing the text here for example in the in in if you notice the curve what do you what we have is a baseline a copy baseline which basically by increasing the noise we notice that the
تعتمد FlowX على بنية (Microservices)، وتشمل خدمة API مبنية بـASP.NET Core لإدارة سير العمل، ومنفذ LangChain بلغة بايثون لتنفيذ عمليات LLM، وواجهة أمامية مبنية بـReact Native لتوفير تجربة تطوير سهلة.
2023.acl-long.809
the problem becomes more and more difficult for the baseline while the normalization models can deal with it in a good way. Our models dramatically improve over the baseline, including when evaluated on real data. So one of the points that we got
تم تصميم نموذج نطاق (Domain Model) يدعم إدارة المستخدمين ووظائف الروبوت، ويستخدم هيكل رسومي يميّز بين روابط التدفق (للتحكم في تسلسل المحادثة) وروابط البيانات (لنقل المعلومات بين العقد).
2023.acl-long.809
The criticism that we got from the reviews was that we did not run our experiments on genuine data. That's why we extended our study on real data. We annotated real data from people who write unconventionally.
يدعم النظام إدارة حالة المحادثة ويوفر تجربة تطوير متكاملة من إنشاء سير العمل، إلى التخصيص المرئي، إلى الاختبار المباشر.
2023.acl-long.809
And notice that basically our normalization models in the last columns improve that performance dramatically, even though there's room for improvement. As you see, 20 points of bleu is still far-fetched.
كما تم تحسين الأداء لضمان عرض فعال لسير العمل المعقد واستجابة سريعة في تفاعلات الروبوت.
2023.acl-long.809
from 100%, right? So we have still some work to do. We also extend our experiments to some downstream tasks like language identification. What we notice is that when we run, let's say a model, an existing model on language identification,
نعتمد أيضًا على تقنيات مثل LangChain لتنفيذ المهام وتجريد التعقيدات التقنية، مما يقلل الحاجة إلى الصيانة من جانب المستخدم.
2023.acl-long.809
to identify text written in an unconventional way, based on different levels of noise, as you notice, from 0 to 100%. By increasing noise, the more the text becomes noisier, we notice that the models
لاحظنا أن الحلول الحالية تنقسم إلى فئتين رئيسيتين: أدوات البناء المرئية مثل Chatfuel وManyChat التي تركز على التدفقات المنظمة لكنها تفتقر إلى قدرات الذكاء الاصطناعي المتقدمة، ومنصات الذكاء الاصطناعي مثل Dialogflow التي تتطلب خبرة تقنية.
2023.acl-long.809
the language identification model struggles to detect the language correctly but when once we apply our normalization model we notice that using the normalization model and normalizing the
FlowX يملأ هذه الفجوة عبر توفير أداة بناء مرئية متاحة تدمج بين النهجين بسلاسة.
2023.acl-long.809
text improves the language identification task dramatically. As you see, it's usually around 0.9 F1 measure. We also extend our work to machine translation. In this task, we translate
بخلاف المنافسين، يقدم FlowX روبوتات هجينة تجمع بين القدرة على التحكم في المحادثة عند الحاجة والاستفادة من نماذج اللغة للتعامل مع المدخلات غير المتوقعة، وذلك ضمن بيئة تطوير موحّدة.
2023.acl-long.809
using the NLLB, no language left behind model. And we tried to see what is the blue score for clean data. These are the results for clean data. And on the other hand, when we have noisy data for Sindhi or Serani Kurdish
باختصار، FlowX يعيد تعريف تطوير روبوتات الدردشة من خلال نهجه الهجين المبتكر.
2023.acl-long.809
Kashmiri, we noticed that the machine translation model cannot successfully translate the sentences. By applying the normalization models, again, similar to the language identification task, we noticed the translation model
فهو يمكّن المستخدمين من تصميم سير عمل مرئي يجمع بين المنطق المنظم والذكاء الاصطناعي الديناميكي، والتكامل مع الخدمات الخارجية عبر طلبات API لأتمتة المهام، وحتى تخصيص مظهر الروبوت، مما يتيح إنشاء روبوتات دردشة من الجيل القادم تتميز بالقوة وسهولة الاستخدام.
2023.acl-long.809
can detect, can translate more accurately. Our conclusions are unconventional writing is more widespread than thought of. It is an open problem. It's non-trivial. And we are far from solving this problem.
تحتوي اللغة العربية على عدد محدد من مجموعات الحروف (particles)، مثل: حروف الجر، وحروف الجزم، وحروف النصب، الخ. وتؤدي كل مجموعة دوراً معيناً في السياق الذي ترد فيه.
2023.acl-long.809
It negatively affects NLP for low-resource languages. And we conclude by asking this question, do we always need to write a language, given that a lot of languages don't have a written form? To what extent we can use the
بشكل عام، فإن الوسم هو عملية تحديد الصيغة المناسبة لكل كلمة سواء أكانت اسماً أو فعلاً أو حرفاً، وذلك بالاعتماد على السياق الذي ترد فيه.
2023.acl-long.809
state-of-the-art techniques or propose new ones to be able to process these languages effectively? Can we use probably multi-model NLP by considering audio or multimedia from these languages or multi-lingual NLP, multi-task learning?
وكذلك يستخدم لتحليل النص لفهم المعنى المطلوب لاستخدامه في محركات البحث (search engines).
2023.acl-long.809
or maybe better adaptation in the current NLP technologies. Our models and codes are available on GitHub and I hope that you enjoyed this presentation. Thank you very much for your attention.
تشمل تطبيقات معالجة اللغات أيضاً الترجمة الآلية (machine translation)، تأليف الكلام (speech synthesis)، التعرف على الكلام (speech recognition)، التشكيل الآلي (diacritization)، وغيرها.
2023.acl-long.23
Hi, I welcome you to a short primer on our work. What about M? How commercial machine translation fades
حوسبة اللغة: ما بين هيمنة الشبكات العصبية وتراجع نماذج ماركوف المخفية
2023.acl-long.23
in a new pronouns. I'm Anna Lauscher and this is joint work with Deborah Notzer, Archie
منذ اكتشافها في بدايات القرن الماضي، حظيت نماذج (ماركوف) على قبول واستحسان الباحثين، واستخدمت في مجالات متعددة.
End of preview. Expand in Data Studio
README.md exists but content is empty.
Downloads last month
1,128