Facebook 活動事件擷取系統(Facebook Activity Event Extraction System)[In Chinese]
نویسندگان
چکیده
The popularity of social networks has made them a perfect medium for activity or advertising campaign promotion. Therefore, many people use Facebook pages to announce their advertising campaign. The purpose of this study is to extract activity events by constructing two named entity recognition models, namely activity name and location, via a Web NER model generation tool [1]. We enhance the tool by improving the tokenizer and alignment technique. In addition, we also use a large database of FB checkin places for location name recognition improvement. For entity relation extraction, we apply sequential pattern mining to The 2016 Conference on Computational Linguistics and Speech Processing ROCLING 2016, pp. 229-243 The Association for Computational Linguistics and Chinese Language Processing
منابع مشابه
基於卷積類神經網路之廣播節目音訊事件偵測系統 (Automatic Audio Event Detection of Broadcast Radio Programs Based on Convolution Neural Networks) [In Chinese]
廣播電臺節目中通常包含語音,音樂與其他音訊事件(如笑聲或特效聲)。若能偵 測並切割這些音訊事件,就能進一步對廣播節目進行加值運用。例如,轉寫語音片段的 逐字稿,或是辨認音樂片段的歌名與曲名,以利檢索。針對此問題,在本論文中,我們 首先設計,並以人工標註出一廣播節目音訊事件資料庫,再利用 Convolutional Neural Network (CNN)自動擷取有效的特徵音訊參數,對廣播電臺的音檔做音訊事件偵測與切 割,最後轉成具時間資訊的音訊事件標註檔。實驗方面我們從教育電臺節目中,選出新 聞類與不同性質的談話類節目共 14 個,經人工標注後,獲得總長度共約 60 小時的音 檔,並用來訓練與測試 CNN和傳統 Gaussian Mixture Model(GMM)的效能。實驗結 果顯示以 CNN直接搭配頻譜參數,在偵測語音與非語音,音樂與非音樂或其它與非其 它音訊事件等的錯誤率(e...
متن کاملJoint Modeling of Trigger Identification and Event Type Determination in Chinese Event Extraction
Currently, Chinese event extraction systems suffer much from the low quality of annotated event corpora and the high ratio of pseudo trigger mentions to true ones. To resolve these two issues, this paper proposes a joint model of trigger identification and event type determination. Besides, several trigger filtering schemas are introduced to filter out those pseudo trigger mentions as many as p...
متن کاملJoint Modeling for Chinese Event Extraction with Rich Linguistic Features
Compared to the amount of research that has been done on English event extraction, there exists relatively little work on Chinese event extraction. We seek to push the frontiers of supervised Chinese event extraction research by proposing two extension to Li et al.'s (2012) state-of-the-art event extraction system. First, we employ a joint modeling approach to event extraction, aiming to addres...
متن کامل基於深層類神經網路之音訊事件偵測系統(Deep Neural Networks for Audio Event Detection)[In Chinese]
現實生活中常有許多聲音事件會一起發生,而聲音會重疊在一起,使得傳統(Gaussian Mixture Model ,GMM)方法很難準確辨認這些重疊的聲音事件。因此,本文提出以深層 類神經網絡(Deep Neural Network, DNN)來檢測這些互相干擾的聲音事件,並據此參加 Detection and Classification of Acoustic Scenes and Events 2016 (DCASE2016) 比賽, DCASE2016 評比提供的音訊資料,內有兩種場景,包括居家與戶外,共有 18 種含有背 景的聲音事件。實驗結果顯示使用 DNN 與傳統 GMM 比較,其場景偵測錯誤率可從 0.91 降至 0.86、F1 分數並從 23.4%提升到 26.8%。此外針對室內環境的音訊事件偵測,錯誤 率可從 1.06 降至 0.86,F1 分數並從 8.9%提升...
متن کاملSemantic Associative Topic Models for Information Retrieval
主題模型(topic model)被廣泛地應用在各種文件建 模以及語音識別、資訊檢索和本文探勘系統中,有 效地擷取文件或字詞的語意和統計資料。大多數主 題模式,例如機率潛在語意分析(probabilistic latent semantic analysis) 和 潛 在 狄 利 克 里 分 配 (latent Dirichlet allocation),主要都透過一組潛藏的主題機 率分布來描述文件與字詞之間的關係,並用以擷取 文件的潛在語意資訊。然而,傳統的主題模型受限 於詞袋(bag-of-words)的假設,其潛藏主題僅能用來 擷取個體詞(individual word)之間的語意資訊。雖然 個體詞可傳達主題信息,但有時會缺乏本文準確的 語意知識,容易造成文件的誤判,降低檢索的品 質。為了改善主題模型的缺點,本論文提出一種新 穎的語意關聯主題模型(semantic associ...
متن کامل