1. 程式人生 > 其它 >Natural Language Question Answering over RDF — A Graph Data Driven Approach論文學習

Natural Language Question Answering over RDF — A Graph Data Driven Approach論文學習

研究內容

  • In this paper, we propose a systematic framework to answer natural language questions over RDF repository (RDF Q/A) from a graph data-driven perspective. (在本文中,我們提出了一個系統的框架,從圖資料驅動的角度來回答RDF儲存庫(RDF Q/ a)上的自然語言問題。

技術方案

  • Generally speaking, there are offline and online phases in our solution.(一般來說,在我們的解決方案中有離線和線上階段。
    • In the offline processing, we propose a graph mining algorithm to map natural language phrases to top-k possible predicates (in a RDF dataset) to form a paraphrase dictionaryD, which is used for question understanding in RDF Q/A.(在離線處理中,我們提出了一種圖挖掘演算法,將自然語言短語對映到 top-k 可能的謂詞(在 RDF 資料集中)形成釋義詞典 D,用於 RDF Q/A 中的問題理解。
    • In the online processing, we adopt two-stage approach. In the query understanding stage, we propose a semantic query graph to model the query intention in the natural language question in a structural way. Then, RDF Q/A is reduced to subgraph matching problem in the query evaluation stage. We resolve the ambiguity at the time when matches of the query are found. The cost of disambiguation is saved if there are no matching found.(在線上處理中,我們採用兩階段的方法。 在查詢理解階段,我們提出了一個語義查詢圖,以結構化的方式對自然語言問題中的查詢意圖進行建模。 然後,RDF Q/A 在查詢評估階段被簡化為子圖匹配問題。 我們在找到查詢匹配項時解決歧義。 如果找不到匹配項,則可以節省消歧成本。
    • We take a lazy approach and push down the disambiguation to the query evaluation stage. (我們採用一種懶惰的方法,將消歧向下推到查詢評估階段

如何做實驗

  • We compare our method with one state-of-the-art algorithm DEANNA and all systems in QALD-3 competition on DBpedia RDF dataset.(我們在 DBpedia RDF 資料集上將我們的方法與一種最先進的演算法 DEANNA 和 QALD-3 競賽中的所有系統進行比較。
  • To build the paraphrase dictionary, we use two different relation phrase datasets in Patty system, wordnet-wikipedia and freebase-wikipediawe.(為了構建釋義詞典,我們使用了Patty系統中兩種不同的關係短語資料集。