TW473675B - Topics dialog procedures and system - Google Patents
Topics dialog procedures and system Download PDFInfo
- Publication number
- TW473675B TW473675B TW089119845A TW89119845A TW473675B TW 473675 B TW473675 B TW 473675B TW 089119845 A TW089119845 A TW 089119845A TW 89119845 A TW89119845 A TW 89119845A TW 473675 B TW473675 B TW 473675B
- Authority
- TW
- Taiwan
- Prior art keywords
- dialogue
- information items
- program
- item
- topic
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2452—Query translation
- G06F16/24522—Translation of natural language queries to structured queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/242—Query formulation
- G06F16/243—Natural language query formulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- User Interface Of Digital Computer (AREA)
Abstract
Description
B7 五、發明説明(i ) 【發明之技術領域】 本發明係有關於一種用於電腦與人間以進行對話之主 題對話方法及系統,特別是一種使用具特有構造之資料 庫,而可就特定話題進行如人與人間自然且理智之對話般 者。 【發明之背景】 近-年來,電腦之聲音辨識技術曰趨發達,電腦與人以 聲音進行對話之環境亦逐漸成熟。因聲音為人類最自然之 輸出入方法’使用儲存於電腦中之資訊而使電腦與人 可如人們錢般進行理智對話之實驗及研究乃甚為盛行。 然而,現在的技術僅限於可由健存於電腦之資訊中檢索並 取出合適者’而單純地將資訊傳達予人,<,配合記錄裝 置中之資料構造,以大致自然之速度正確地進行詢問或回 答。 , 前者之代表例有諸如日本公開公報特開平U-272747 號。該發明係於需要作業指示之裝配作業場中除一如以 往地對作業者提供設計圖、工作指示書等視覺中心資訊 外’並即時提供諸如各程序之注意事項及過去之不良事例 等之耳語情報,即,於電腦中設置内含事先已登錄之訊息 文節的聲音訊息編輯播放裝置,使之可依作業程序,接.受 該事先已登錄之訊息文節的指示,而決定作業指示。 後者之代表例則有諸如日本公開公^'特開平 η-·96號。該發明之系統可於開始對話時,由語言處理 部運用句型模式辭典、任務知键咨ΜB7 V. Description of the Invention (i) [Technical Field of the Invention] The present invention relates to a subject dialogue method and system for dialogue between a computer and a person, especially a database with a unique structure, which can be used for specific purposes. Topics are like natural and rational conversations between people. [Background of the invention] In recent years, computer voice recognition technology has become more developed, and the environment for computers to talk with people through voice has gradually matured. Because sound is the most natural input and output method for human beings, experiments and researches that allow computers and people to conduct a rational conversation like people's money are very popular. However, the current technology is limited to information that can be retrieved from a computer and retrieved from the appropriate person, and simply communicates the information to the person, < cooperates with the data structure in the recording device to make an accurate inquiry at a natural speed Or answer. A representative example of the former is Japanese Unexamined Patent Publication No. U-272747. The invention is to provide the operator with visual center information such as design drawings, work instructions, etc. in the assembly work area that requires work instructions as usual, and provide whisper information such as the precautions of various procedures and past bad cases in real time. That is, a sound message editing and playback device containing a previously registered message section is set in a computer, so that it can receive and receive instructions from the previously registered message section to determine a work instruction. Representative examples of the latter include, for example, Japanese Laid-Open Patent Publication No. JP-A-η- · 96. In the system of the invention, when the conversation is started, the language processing department uses a sentence pattern dictionary and task knowledge keys to consult
•4- 473671• 4- 473671
、發明説明( 經濟部智慧財產局員工消費合作社印製 、句型作成輸出候選句及辨識候選句。然後,由聲土人 成:輸出該等作成之輸出候選句之—。一旦使用者以聲: 應合之,聲音辨識部即優先與以該表達·句型作成之辨識 、選句比較若辨識為正確,則將辨識結果之辨識單詞^ 至wfe處理部,語言處理部即依該辨識結果作成次一輸出 候選句及辨識候選句。若無聲音輸人或未辨識為正確時, 則對應、作成次一輸出候選句等。如此,反覆進行以上程序, 以進行對話。 該糸統雖未明示’但事實上其係利用儲存於人稱任務 知識身料之關連式資料庫中之綱目:與元組而指定話題後, 根據其為6己載於話題資料之詢問或回答等之發話類型話 2及次發話類型(若發話類型為詢問,則次發話類型為回 答’若前者為回答’則後者為詢問)’而由句型模式辭典選 出預定之模式(句型)者。模式係視發話類型而選出符合現 在之對話場面者,而模式之預定位置上則設置任務知識資 料所對應之方案之元組。該系統除係以如上方式作成輸 出候選句(輸出候選單詞)外,並具有如下特徵,即同時作 成與輸出候選句相同模式號碼之辨識候選句。此處,並於 相同模式號碼相對應地標註同一文章之詢問與回答。因 此’所謂準備相同模式號碼之輸出候選句與辨識候選句即 意指事先準備回應—個問題或回答之可能性最高的文章。 後’將該等輸出候選句與辨識候選句送至聲音辨識部及 聲音合成部。 聲音合成部係可使輸出候選句之一以聲音 私衣IX------0 (請先閱讀背面之注意事項再填寫本頁)Description of the invention (Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, the sentence pattern is used to generate output candidate sentences and identification candidate sentences. Then, the native person is to output the output candidate sentences that are made— once the user speaks : As a coincidence, the voice recognition unit will prioritize recognition and sentence selection made with the expression and sentence pattern. If the recognition is correct, the recognition word of the recognition result ^ will be sent to the wfe processing unit, and the language processing unit will create the recognition result based on the recognition result. The next output candidate sentence and recognition candidate sentence. If there is no voice input or the recognition is not correct, then the next output candidate sentence is correspondingly made, etc. In this way, the above procedure is repeatedly performed for dialogue. Although this system is not explicitly stated 'But in fact, it uses the outline stored in the related database of personal task knowledge and body: after specifying the topic with the tuple, it is based on 6 types of speeches such as inquiries or answers contained in the topic information. 2 The type of secondary speech (if the speech type is inquiry, the secondary speech type is to answer 'if the former is an answer', then the latter is an inquiry) 'and a predetermined one is selected from the sentence pattern dictionary The pattern (sentence pattern). The pattern is selected according to the type of speech, and the tuple corresponding to the task knowledge data is set at the predetermined position of the pattern. In addition, the system creates output candidates in the above manner. In addition to sentences (output candidate words), it also has the feature of simultaneously creating recognition candidate sentences with the same pattern number as the output candidate sentence. Here, the questions and answers of the same article are marked correspondingly with the same pattern number. Therefore, the so-called Preparing output candidate sentences and recognition candidate sentences with the same pattern number means to prepare in advance a response to an article with the highest probability of answering a question or answer. Then, send these output candidate sentences and recognition candidate sentences to the voice recognition department and voice synthesis. The sound synthesis department can make one of the output candidate sentences with voice private IX ------ 0 (Please read the precautions on the back before filling this page)
-5- 4736?(〇 五 經濟部智慧財產局員工消費合作社印製 、發明説明(-5- 4736? (〇 5 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, Invention Description (
I 用者輸出者。若使用者未對該輸出句輸入回應之聲音,則 :返:輸出候選句等之作成程序,以作成次一輸出候選 句二若有聲音輸入,則將以模式號碼與輸出候選句相同之 辨識候選句為優先而與使用者所輸入之聲音相比較。輸入 之聲音與辨識候選句是否—致之騎料藉單詞分解及/ 或文章解析而輕易進行。如上所述,當聲音賴部對照辨 識候選^句與輸入之聲音而判斷辨識結果為正確時,即將辨 識早河达至語言處理部。語言處理部則於藉該辨識結果完 成任務時結束該任務。 核對辨識候選句與輸入之聲音後,若判斷輸入之聲音 與#識候ϋ句並不—& ’則再對照其他辨識候冑句,並反 訂 覆進行該作業直至尋獲與輸入之聲音一致的辨識候選句為 止。 乍看上述電腦與人之對話,好似透過資訊之交流而進 行著理智的對話。然而,事實上僅止於單方面對人傳達電 線 腦中所儲存之資訊而已。要不是,僅止於可在短時間内對 人所送出之資訊正確地加以辨識並予以資料庫化而已;便 疋’僅止於檢索依以上程序資料庫化後之資訊並予以輸出 而已。 人與人間之對話,即便是同為處理資訊之對話,亦•不 致一次突然處理眾多冗長資訊,而通常會分成彼此容易理 解之長度、項數’而依其重要性由高至低或由相關性較高 者依次展開對話,並一面加以歸納等。其次,檢索資訊後 若有欠缺之項目時,則待適當時機提出徵詢,臨機應變。 •6· 本紙張尺度適用中國國家標準(CNS ) Α4規格(210X297公釐) 47367{〇I user exporter. If the user does not input a response voice to the output sentence, then: return: output candidate sentence creation procedures to create the second output candidate sentence. If there is a voice input, the pattern number is the same as the output candidate sentence. The candidate sentence is given priority and compared with the sound input by the user. The input voice and recognition of the candidate sentence are all made easy by word decomposition and / or article parsing. As described above, when the voice reliance section judges that the recognition result is correct by comparing the recognition candidate sentence with the input voice, the recognition is soon to reach the language processing section. The language processing unit ends the task when the task is completed based on the recognition result. After checking the recognition candidate sentence and the input voice, if it is judged that the input voice and #cognition haiku are not— & ', then check the other recognition haikus, and repeat the operation until the voice is found and input. Candidates are identified consistently. At first glance, the computer-to-human dialogue above seems to be a rational dialogue through the exchange of information. However, the fact is that the information stored in the brain of the wire is only unilaterally conveyed to the person. If not, it is only that the information sent by a person can be correctly identified and databased in a short period of time; it is only limited to retrieving and outputting the information that has been databased according to the above procedure. People-to-people dialogues, even if they are also dealing with information, do not suddenly process a lot of lengthy information at one time, but are usually divided into lengths and items that are easy to understand each other, and they are ranked according to their importance from high to low or related The sexiers start the conversation in turn and summarize it on the side. Secondly, if there are any missing items after retrieving the information, it should be consulted at an appropriate time and the situation should be adjusted. • 6 · This paper size applies to China National Standard (CNS) Α4 size (210X297 mm) 47367 {〇
另,當一資訊係由多個項目所構成時,若直接且連續地徵 詢必要之項目,將予人機械化或公事化之感,而不能稱之 為具人情味的對話。此時即應對乍看似無關連之方向途徑 或略寅關連之項目之相關資訊深入加以探索,一面繼續進 行對話,以取得所需項目之相關資訊,依此方式進行而 在最後方才取得檢索所需之全部條件。本類技術之目標即 在使電·腦與人之對話中亦可進行此等自然且理智之對話。 因此,此等主題對話方法及系統之開發,長久以來已為人 們之所熱切期許。 而,本發明即係因應該期許而設計者,其目的在提供 一種可於電腦與人間進行自然且理智之對話之主題對話方 法及系統。 本發明之目的亦在提供一種可藉電腦置換及利用既有 之關連式資料庫及物件導向資料庫中之資訊而作成知識化 為料,並以之為基礎而與人進行對話之主題對話方法及系 統。 本發明之目的並在提供一種可使電腦自動搜尋欠缺之 項目,而於進行自然會話時取得該項目之相關資訊,以於 儲存之資料構造中進行檢索之用於電腦與人間之簡易主題 對話方法及系統。 本發明之目的更在提供一種可對乍看似無關連之方向 途徑或有關之項目之相關資訊深入加以探索,一面繼續進 仃對話,最後取得必要項目之所有相關資訊以進行檢索及 資訊儲存,而非進行直接連續詢問必要之項目之所謂機械 〔锖先閱讀背t&意事項真填菸本真) 装·In addition, when an information is composed of multiple items, if the necessary items are directly and continuously consulted, it will give people a sense of mechanization or officialization, and cannot be called a human dialogue. At this point, you should explore the relevant information of the seemingly unrelated direction or slightly related projects, and continue the dialogue to obtain the relevant information of the required projects, and proceed in this way to obtain the search office at the end. All the conditions needed. The goal of this type of technology is to enable such natural and rational dialogues in the dialogue between electricity, brain, and people. Therefore, the development of these thematic dialogue methods and systems has long been eagerly anticipated. However, the present invention is designed as expected, and its purpose is to provide a thematic dialogue method and system capable of natural and rational dialogue between computers and people. The object of the present invention is also to provide a thematic dialogue method that can be used to replace knowledge and information in existing related data bases and object-oriented data bases with computers, and use them as a basis for dialogue with people. And system. It is an object of the present invention to provide a simple theme dialogue method for a computer and a person, which can enable a computer to automatically search for a missing item and obtain relevant information of the item during natural conversation for retrieval in a stored data structure. And system. The purpose of the present invention is to provide an in-depth exploration of the related information of the seemingly irrelevant direction or related items, while continuing the conversation, and finally obtaining all the relevant information of the necessary items for retrieval and information storage, Rather than the so-called machinery that directly and continuously inquires about the necessary items (read the t & suggestions first and fill the cigarettes really)
1T 經濟部智慧財產局員工消費合作社印製 •7- 473671) A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明説明( 化或公事化對話之用於電腦與人間之主題對話方法及系 統。 【發明之概要】 本發明之第一樣態可提供一種主題對話方法,係用於 電腦與人間自然且理智之主題對話者,包含有:一記憶體 記錄程序,係於電腦之記憶體中記錄關連式資料庫之綱目 及元纽;·,同時並記錄設定了以各方案為話題時之次序的對 話順序及各方案成為話題後之措辭等之程式者;一判斷程 序,係對使用者所輸入之聲音進行單詞分解及/或文章解 析’以判斷有無與各方案相對應之資訊項者;一徵詢程序, 係於輸入之聲音中未包含用以指定關連式資料庫中之記錄 所需之資訊項時,就欠缺之資訊項徵詢使用者,以取得該 所而之全部資訊項目者;及,一程式進行程序,係於輸入 之聲音中包含有用以指定記錄所需之資訊項時,使用該記 錄之資訊項,依照對話順序使程式進行者。 在本樣態中’電腦係使用儲存於關連式資料庫中之資 料與人類對話。該關連式資料庫中儲存有眾多方案,即, 與綱目相對應而儲存有資料(元組)。如此之關連式資料庫 係記錄於電腦之記憶體中。又,記憶體中並記錄有在與人 類對話時決定以何順序將各方案加入話題之對話順序與•用 以規定各個方案成為話題時之措辭、其變形等之程式。電 腦之CPU則可由記憶體叫出該程式而依照該程式繼續進行 對話。 一旦人類對電腦發出聲音,電腦即運用麥克風、聲音 &張尺反·中關家標準(CNS) A4規格(21GX297公瘦) ---------装------、玎------線 (請先閲讀背面之注意事項再填寫本頁} •8· 經濟部智慧財產局員工消費合作社印製 4736?〇 A7 ^^ _____Β7__ 五、發明説明(6 )~一~' - 辨識軟體等以辨識音素。利用單詞辭典、構句辭典、事例 辭典等以對其進行單詞分解及/或文章解析。其次,判斷與 各方案相對應之值(此處稱為「資訊項」)是否存在於業經 單詞分解及/或文章解析之輸入之聲音中。關連式資料庫中 儲存有眾多記錄,而該等記錄則係用以決定輸入之聲音與 其中何記錄相關者。即,當輸入之聲音中未包含用以指定 圮錄所_需之資訊項時,則就欠缺之資訊項徵詢使用者,以 取得該所需之全部資訊項目。藉此,即可指定一筆或數筆 記錄。 當用以於業經單詞分解·文章:解析之輸入之聲音中指 定記錄之資訊項自始即完全齊備時(包含上述進行徵詢使 用者後而齊備之情況。以下亦相同。),電腦之cpu即可使 用該記錄之資訊項並依照對話順序使程式進行。通常,在 對話順序中皆使用與該記錄之全部或合適之方案相對應之 資訊項而依照預定之順序使對話進行。 本發明之申请專利範圍第2項係如申請專利範圍第j項 之主題對話方法,其中該欠缺之資訊項之徵詢程序係將方 案名稱加入徵詢句以進行詢問者。 使用者可由徵詢句中之「方案名稱」而正確地把握被 詢問之問題,而確實取得意向之溝通。 本發明之第二樣態可提供一種主題對話方法,係用於 電腦與人間自然且理智之主題對話者,包含有:一記憶體 記錄程序,係於電腦之記憶體中記錄具有可識別各情境(話 題)之目錄之諸多知識化資料之屬性及等級,以以 本紙張尺度適用中國國家標準(CNS ) A4規格(210x297公瘦) 裝------訂------線 (請先閲讀背面之注意事項再填寫本頁) -9- 473671) A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明説明( 規定與全部或預定之情境之預定方案及/或元組相關連之 小情境之關連式資料庫之諸多項目所構成之構造事例及其 内容之内容事例,同時記錄設定了以各方案及項目為話題 時之次序的對話順序及項目順序,以及,各計劃及項目成 為話題後之措辭等之程式者;一詢問程序,係由電腦向人 詢問所欲之目錄,以指定其為何種情境之對話者;一快取 δ己憶體J己憶程序’係依照輸入之目錄指定情境,再由記憶 體叫出所指定之關連式資料庫而將之記錄於快取記憶體 中’並使其呈可自由改寫之狀態者;一判斷程序,係對使 用者所輸入之聲音進行單詞分解及/或文章解析,以判斷 有無與各方案相對應之資訊項者;一徵詢程序,係於輸入 之聲a中未包含用以指定關連式資料庫中之記錄所需之資 訊項時,就欠缺之資訊項徵詢使用者,以取得該所需之全 部資訊項目者;一程式進行程序,係於輸入之聲音中包含 有用以指定記錄所需之資訊項時,使用該記錄之資訊項, 依照對話順序使程式進行者;一次常式進行程序,係於預 定方案及/或元組成為話題時,叫出用以規定小情境之關連 式資料庫以作為次常式,而依照項目順序使程式進行之程 序,及,一餘式進行程序,係於項目順序結束時,回復為 對話順序而使其餘程式進行者。 < 本樣態係有關於一種可於電腦與人類間進行更複雜而 T智之對話之主題對話方法。電腦係使用具有可識別各情 境(話題)之目錄(亦可由方案及/或元組中選擇合適者加以 使用。以下亦相同。)之諸多關連式資料庫以與人類對話。 &張家標準(CNS) A4規格(]〇χ297公董 --------^--种衣------1T------0 (請先閲讀背面之注意事項再填寫本頁) -10- 47367b 五、發明説明(8 經濟部智慧財產局員工消費合作社印製 該等關連式資料庠中分別儲存有I多方案,即,與綱目相 對應而儲存有資料(元組)。具有可進行識別之目錄之關連 式#料庫則係記錄於電腦之記憶體中。又,記憶體中並記 錄有^以規^與所有情境或預s之情境之預定;r案及/或 元、、且相關之情境之關連式資料庫。用以規定子情境之關 ,式資料庫係由諸多項目所構成之構造事例與其内容之内 谷事例·所構成者。且’記憶體中並記錄有在與人類對話時, 決疋以何-人序將各方案及項目加入話題之對話順序及項目 貭序以及β又疋了各方案及項目成為話題時之措辭、變形 等之耘式。電腦之CPU則可由記憶,趙叫出該程式而依照該 程式繼續進行對話。 程式之開頭係由CPU向人詢問所欲之目錄,以指定其 為何種情境之對話,例如·· Γ請問您需要什麼服務?」、「您 想查詢的是?」等。-旦人類對此作出「我要找暑假的旅遊 去處。」等回答,則依照「旅行去處」、「找」等輸入之 目錄而指定情境。CPU則由記憶體叫出依目錄而指定之情 境(話題)之關連式資料庫而將之記錄於快取記憶體中,並 使其呈可自由改寫之狀態。 對於電腦之「您的旅行計劃為何?」等問題,若使用 加以口合’電知即使用以彳主為人所熟知之聲音辨識裝置 辨識音素。然後利用單詞辭典、構句辭典、事例辭典等對 該業經辨識音素之文字列進行單詞分解及/或文章解析。 旅行情境之綱目有諸如「目的地」、「目的」、 數」、「出發日(時)」、「人數」、「同行者之明細 者 以 曰 装------1T------^ (請先閱讀背面之注意事項再填寫本頁) -11· ^367^ ^367^ 經濟部智慧財產局員工消費合作社印製 A7 ' -------------B7__ 五、發明説明(9 ) 「預算」、「指定夕「> 疋之航I公司」、「扣定之飯店」、「房 門的種類」、「有無附餐」、「有無自費活動」、「有無 護照」、「是否需要簽證」、「付費方式」等。使用者之 回答則通常僅就該等綱目之一部分給予資訊項。CPU即藉 2判斷有無與各方案相對應之資訊項。關連式資料庫中包 含有眾多記錄。為由其中指定符合期望之記錄,則需要有 關數個-方案之資訊項。舉例言之,「目的地」、「目的」、 :日數」、「出發日(時)」、「人數」、「同行者之明細」、 「預算」即為指定記錄所需之資訊項。相對於此,「指定 之航空公司」、「指定之飯店」、「房間的種類」、「有 無附餐」、「有無自費活動」、「有無護照」、「是否需 要簽證」、「付費方式」則為已大略確定旅行之細節後所 需之資訊項。 因此,當輸入之聲音中未包含指定關連式資料庫中之 記錄所需之資訊項時,則就欠缺之資訊項徵詢使用者以取 得該必要之全部資訊項。另,當輸入之聲音中包含有指定 記錄所需之資訊項時,則使用該記錄之資訊項,依照對話 順序使程式進行。通常,會向使用者依序確認輸入完成之 綱目以進行之。舉例言之,當使用者就「目的地」、「目 的」、「出發曰(時)」回答「今年暑假想去美國學英文! 時,即以「目的地是美國嗎?」、「旅行的目的是語言學習 嗎?」、「旅行是從7月20日以後至8月3 1日嗎?」向使用者 確認資訊項是否已正確地為電腦所辨識。 上述使用者之回答中,並不包含有關確定或針對「曰 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) 裝 訂 線 (請先閲讀背面之注意事項再填寫本頁) •12- 4736?{〇 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明説明(10 ) 數」、「人數」、「同行者之明細」、「預算」等記錄所 需之綱目之資訊項。因&,電腦將於上述用於確認之對話 進仃時’將有關該等方案之問題加入以獲得必要資訊項。 舉例•言之’接續「旅行的目的是語言學習嗎?」之問題及使 用者之回荅,再詢問「請問是一個人旅行嗎?」。相對於此, 舉例言之,若回答係「和财共3人_起去。」,則再問「同 行者是-朋友嗎?」以詢問「同行者之明細」。 本樣態之特徵即在於對話當中,提取該對話中出現之 主題,再就該主題深入分析,或言及其他變化等,藉此而 :使對話發展之處。即,當預定之方案及/或元組成為話題 時,可叫出用以規定小情境之關連式資料庫作為次常式, 而依照項目順序使程式進行。在上述之例中,舉例言之, 當使用者言及旅行之「目的」之「學英文」時,即控制之 以將話題轉為關於「語言學習」之小情境。 在「語言學習」之小情境中,則使用由眾多項目所構 成之構造事例及其内容之内容事例所構成之關連式資料庠 以進行與人之對話。對話中項目之登場次序係依項目順序 而疋。語言學習之項目有「語言之種類」、「美英語之別」、 目的」、「特徵」、「場所」、「設施」、「期間」、 形怨」、「學·生之組成」、「費用」、「入學資格」等。 在上述之例中,即以「語言學習有英語與美語。您確實學 過美語嗎?」等,將對話内容轉變為小情境之語言學習之話 題。在該小情境中,則再進一步就日常會話之習得、留學 或移居則之準備、透過曰文教授之英語習得、英語資格考 Z紙張尺度&中關ii^7_CNS) A齡(2igx29 ---------装------、玎------線 (請先閲讀背面之注意事項再填寫本頁) -13- 經濟部智慧財產局員工消費合作社印製 :)之資料中之單詞,依各名稱加以分類,並以作為= 名之名稱與該名稱之類型、大小及可取得之範員1T Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs • 7-473671) A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs [Summary of the invention] The first aspect of the present invention can provide a method of topic dialogue, which is used for natural and rational topic interlocutors between computers and people, including: a memory recording program, which is stored in the computer's memory Record the outline of the relational database and the elements; at the same time, record the dialogue order that sets the order when each plan is the topic, and the formulas of the wording after each plan becomes the topic; a judgment process, which is for users The input voice is subjected to word decomposition and / or article analysis to determine whether there is an information item corresponding to each scheme; a consultation process is not included in the input voice to specify the records in the connected database Information items, consult users for missing information items to obtain all the information items in that place; and, When the input voice contains information items that are needed to specify the record, the recorded information items are used to make the program performers in accordance with the dialogue order. In this aspect, the 'computer uses the data stored in the relational database and Human conversations. The relational database stores many schemes, that is, data (tuples) corresponding to the outline. Such a relational database is recorded in the computer's memory. Moreover, Recorded in the dialogue with human beings is the order in which each plan is added to the topic and the program used to specify the wording of each plan when it becomes the topic, its variants, etc. The computer's CPU can call the program from memory. Continuing the conversation in accordance with the program. Once a human makes a sound on the computer, the computer uses the microphone, sound & Zhang Chi Fang · Zhongguanjia Standard (CNS) A4 specification (21GX297 male thin) --------- install ------ 、 玎 ------ line (Please read the notes on the back before filling out this page} • 8 · Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 4736? 〇A7 ^^ _____ Β7__ V. hair Explanation (6) ~ 一 ~ '-Identification software, etc. to identify phonemes. Use word dictionaries, sentence construction dictionaries, case dictionaries, etc. to perform word decomposition and / or article analysis. Second, judge the values corresponding to each scheme ( (Referred to herein as the "information item") whether it exists in the input voice that has been decomposed by words and / or parsed by the article. There are many records stored in the connected database, and these records are used to determine the input voice and the Who records the relevant person. That is, when the inputted voice does not include the information items required to specify the recording, the user is consulted for the missing information items to obtain all the required information items. One or more records can be specified. When the information items used to designate records in the sound of the word decomposition and article: parsed input sounds are completely complete from the beginning (including the above situation after consulting the user and being complete. The same applies to the following. ), The CPU of the computer can use the recorded information item and make the program proceed according to the dialogue order. In general, the dialogue order is carried out in a predetermined order using information items corresponding to all or appropriate schemes of the record. Item 2 of the scope of patent application of the present invention is the method of topic dialogue as described in item j of scope of patent application, wherein the inquiry procedure of the missing information item is to add the solution name to the inquiry sentence for inquiry. The user can correctly grasp the question being asked by the "program name" in the inquiry sentence, and indeed obtain the intended communication. The second aspect of the present invention can provide a method for thematic dialogue, which is used for natural and rational thematic interlocutors between computers and people, including: a memory recording program, which is recorded in the computer's memory and has identifiable situations (Topics) The attributes and grades of many knowledgeable materials in the catalog are based on the Chinese paper standard (CNS) A4 (210x297 male thin) in this paper size. Please read the notes on the back before filling in this page) -9- 473671) A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs V. Invention Description (Required to be related to all or predetermined scenarios and / or tuples The construction examples and content examples of the contents of the connected database in the connected small-scenario database, and the dialogue order and project order that set the order when each plan and project are set as topics, and each plan and project The item becomes the formulator of the wording after the topic, etc .; an inquiry program, which asks a person to search the desired directory to specify the situation's interlocutor; "Recall procedure" refers to the situation specified according to the input directory, and then the memory calls out the specified relational database and records it in the cache memory "and makes it freely rewriteable; a judgment procedure, Word decomposition and / or article analysis of the voice input by the user to determine whether there is an information item corresponding to each scheme; a consultation process, which is not included in the input voice a to specify a related database When recording the required information items, the user is consulted for the missing information items to obtain all the required information items; a program performs a procedure in which the input voice contains useful information items required to specify the records At the time, the recorded information items are used to make the program performers in accordance with the order of dialogue; once a routine process is performed, when a predetermined plan and / or meta-composition is the topic, a relational database for defining small scenarios is called as Subroutines, and procedures that make the program proceed in accordance with the order of items, and a residual process that reverts to the order of the dialog at the end of the order of the items and makes the rest of the process ≪ This aspect is about a thematic dialogue method that allows more complex and intelligent dialogues between computers and humans. Computers use a catalogue that can identify situations (topics) (also by schemes and / or Choose the appropriate one from the tuple and use it. The same applies below.) Many connected databases to talk to humans. &Amp; Zhang Jia Standard (CNS) A4 Specifications --Seeds ------ 1T ------ 0 (Please read the notes on the back before filling this page) -10- 47367b V. Description of the invention (8 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs The I-type schemes are stored in these related data sets, that is, the data (tuples) corresponding to the outlines are stored. The related #material library with an identifiable directory is recorded in the computer's memory in. In addition, there are recorded in the memory ^ the rules ^ and all scenarios or pre-scenes of the reservation; r case and / or yuan, and related context database. It is used to define the sub-situation, and the type database is composed of the structural examples and contents of many items. And in the memory, it is recorded in the dialogue with human beings, the order of dialogue and project order that decides which programs and projects are to be added to the topic and β, and the wording when each program and project becomes a topic, Deformation and so on. The computer's CPU can be remembered, Zhao calls the program and continues the conversation according to the program. At the beginning of the program, the CPU asks people for the desired directory to specify the context of the dialogue. For example ... What service do you need? "," What are you looking for? " -Once the human beings answered "I'm looking for a travel destination for the summer vacation." And so on, specify the situation according to the input list of "Travel Destination" and "Find". The CPU calls the related database of the situation (topic) specified by the directory from the memory, records it in the cache memory, and makes it freely rewritable. Regarding the computer's "What is your travel plan?" And other questions, if you use the "Matching" message, you can use it to recognize the phoneme by using a sound recognition device known to the owner. Then use word dictionary, sentence formation dictionary, case dictionary, etc. to perform word decomposition and / or article analysis on the character string of the identified phonemes. The outline of the travel situation is such as "destination", "purpose", "count", "departure date (hour)", "number of people", "details of the companion" ------ 1T ---- -^ (Please read the notes on the back before filling out this page) -11 · ^ 367 ^ ^ 367 ^ Printed by A7 of Consumer Cooperatives of Intellectual Property Bureau of the Ministry of Economic Affairs' ------------- B7__ V. Description of the invention (9) "Budget", "Designated evening" > 疋 之 航 I 公司 "," Deducted Hotel "," Type of room door "," With or without meals "," With or without own expense activities ", "Has a passport", "Do you need a visa", "Payment method", etc. The user's response usually gives information items only for a part of the outline. The CPU uses 2 to determine whether there are information items corresponding to each scheme. The connected database contains numerous records. In order to specify records that meet expectations, several information items are required. For example, "destination", "purpose", "number of days", "departure day (hour)", "number of people", "partner details", and "budget" are the information items required for the designated record. In contrast, "designated airlines", "designated hotels", "types of rooms", "with or without meals", "with or without self-funded activities", "with or without passports", "do you need a visa", "payment methods" The items of information required after the details of the trip have been roughly determined. Therefore, when the inputted voice does not contain the information items required for the records in the specified connected database, the user is consulted for the missing information items to obtain all the necessary information items. In addition, when the input voice contains the information items required for the specified record, the recorded information items are used to make the program proceed in accordance with the dialogue order. Normally, the user will sequentially confirm the input to complete the program to proceed. For example, when the user answers "Destination", "Purpose", "Departure (time)", "Want to go to the United States to learn English this summer!", "Is the destination a US?" Is the purpose of language learning? "," Is the trip from July 20 to August 31? "Confirm with the user whether the information item is correctly recognized by the computer. The above user ’s answer does not include the determination or application of “Chinese National Standard (CNS) A4 Specification (210X297 mm) for Japanese paper sizes” gutter (please read the precautions on the back before filling out this page) • 12 -4736? {〇 A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 5. Inventory (10) Numbers "," Number of People "," Details of Travelers "," Budget "and other information items required for the outline of records . Because of & the computer will add questions about these programs to obtain the necessary information items when the dialogue for confirmation mentioned above is advanced. For example, the words “continue the question of“ Is the purpose of travelling in language learning? ”Followed by the user ’s response, and then ask“ Is this travelling alone? ”. On the other hand, for example, if the answer is "3 people together with Choi Choi.", Then ask "Is the companion a friend?" To ask "Details of companion." The characteristic of this aspect lies in the dialogue, extracting the theme appearing in the dialogue, then in-depth analysis of the theme, or talking about other changes, etc., so as to make the dialogue develop. That is, when a predetermined scheme and / or meta-composition is a topic, a relational database for specifying a small context can be called as a sub-routine, and the program can be performed according to the order of items. In the above example, for example, when the user talks about the “learning English” of the “purpose” of travel, it is controlled to turn the topic into a small context of “language learning”. In the small context of "Language Learning", the use of related materials composed of construction examples and content content examples composed of many projects for dialogue with people. The order of appearance of the items in the dialogue is based on the order of the items. Language learning items include "types of language", "differences between American and English", purpose "," characteristics "," place "," facilities "," periods ", resentment," composition of learning and students "," Fees, "" eligibility, "and so on. In the above example, "Language learning has English and American language. Have you ever learned American language?", Etc., to transform the content of the conversation into a small context of language learning. In this small situation, you will further acquire the daily conversation, the preparation for studying abroad or moving, the English acquisition through Professor Yuewen, the English qualification test Z paper scale & Zhongguan ii ^ 7_CNS) A age (2igx29- ------- install ------, ---------- line (please read the precautions on the back before filling this page) -13- Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs: The words in the data are classified according to each name, and the name as = name and the type, size, and available members of the name
本紙張尺度朝家轉(CNS ) 21GX297公釐 473671) 五、發明説明(11 ) 試之準備等有關「目地」之項目、使用大學校舍進修、银 羅-流之教師群、寓教於樂之學習等有關「特徵 曰 以及在什麼國家之何處進行之所謂有關「場所」之項目、料 繼續進行對話。 、目寺 就小情境之關連式資料庫中之構造事例之全 r完成項目順序時,即回復為對話順序而使其餘程式: 本發明之巾請專利·第4㈣如t請專利範圍第3項 之主㈣話方法,其中該構造事例係指對話例。過、 情境之對話例係依照關連式資料庫之構造事例之對話^ 係可如常作為記錄而儲存於該關連式資料庫。因此二 =過去之該對話例作為構造事例,可將相似之例子作二 本發明之申請專利範圍第5項係如 ^項之主題對話方法,其中該小情境並具有居其下位第之't 又’該居於下位之小情境又具有更居其τ位之 *兄’而形成無限之階層者。 :、情境所形成之會話之次常式可形成多級階層藉 此,即可進行富於變化且更具發展性之對話。 κ本發明之第3樣態可提供-種主題對話方法,係用‘電 月®與人間自然且理智之主題對話者八 、 詞辭典作成程序,係U二The size of this paper is for home (CNS) 21GX297 mm 473671) 5. Description of the invention (11) Preparation of the test and other items related to the "purpose", the use of university school for further studies, the group of teachers of Yinluo-stream, and the joy of teaching We will continue the dialogue on the so-called "places" related to "characteristics, and in what countries and where". When Mu temple completes the project sequence for all the construction examples in the small-scenario relational database, it reverts to the dialogue sequence and makes the rest of the program: the patent for the present invention, and the fourth item, such as the third item in the patent scope. The main method of speaking, in which the construction case refers to a dialogue case. However, the dialogue example of the situation is a dialogue according to the construction example of the relational database ^ It can be stored as a record in the relational database as usual. Therefore, the second = the past example of dialogue is used as a construction example. A similar example can be used as the second topic of the present invention. The scope of the patent application is subject to the method of dialogue, where the small context has the lowest rank. Those who should be in the lower position have the "brother" who is in the lower position and form an infinite class. : The sub-routine of the conversation formed by the situation can form a multi-level hierarchy, which can carry out a dialogue that is rich in change and more developmental. κ The third aspect of the present invention can provide a method of thematic dialogue, which is a program using ‘Dian Yue® and the natural and rational thematic interlocutor VIII, thesaurus dictionary, U2
d------、玎! (請先閱讀背面之注意事項再填苟本頁) 線 •14· 4736?ϋ 經濟部智慧財產局員工消費合作社印製 五、發明説明(l2 ) .、屬性,且以該名稱之值作為等級之知識化單詞加以累 積而作成知識化單詞辭典者;一記憶體記錄程序,係於 電腦之記憶體中記錄包含相同屬性組之多個或單個知識化 貝料瘁所具有之諸多屬性/等級組,同時記錄設定了以各屬 性為話題時之次序的對話順序及各屬性成為話題後之措辭 等之程式者,一判斷程序,係利用知識化單詞辭典對使用 者所輸-入之聲音進行單詞分解及/或文章解析,以判斷有 無與各屬性相對應之資訊項者;一徵詢程序,係於輸入之 聲曰中未包含用以指定知識化資料所需之資訊項時,就欠 缺之資訊項徵詢使用者,以取得該所需之全部資訊項目 者,及 私式進行程序,係於輸入之聲音中包含用以指 定知識化資料所需之資訊項時,使用該知識化資料之資訊 項,依照對話順序使程式進行者。 本樣態係使用了具通用性並可正確而迅速地辨識輸入 之聲音之知識化資料庫以替代上述第一樣態中所使用之關 連式資料庫者。因此,在本樣態中,電腦係使用儲存於知 識化資料庫中之知識化資料而與人對話。 欲作成該知識化資料庫時,首先,須將出現於屬預定 對話所進行之情境(話題)之資料中之單詞依各名稱加以分 類。有關旅行之情境可使用有關旅行之報導、書、綱路之 綱頁、小冊子及其他資料。屬性則包含有作為分類名之名 稱與該名稱之類型、大小及可取得之範圍等條件。在上述 之例中,「目的地」即出現了「文字」之「固有名詞」或 「表示場所之名詞」。夏威夷、瑞士、&美、盧布館 (請先閱讀背面之注意事項再填寫本頁) -裝· -ue 線d ------, oh! (Please read the notes on the back before filling this page) Line • 14 · 4736? 印 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (l2), attributes, and the value of the name as the level Those who accumulate knowledgeable words to make a dictionary of knowledgeable words; a memory recording program that records in a computer's memory multiple or single knowledgeable materials containing the same attribute group At the same time, recorders who set the dialogue order that sets the order of each attribute as the topic and the wording after each attribute becomes the topic. A judgment procedure is to use the knowledgeable word dictionary to word the input-input voice of the user. Decomposition and / or article analysis to determine whether there are information items corresponding to each attribute; a consultation process is the lack of information when the input voice does not include the information items required to specify the knowledgeable data The user is consulted to obtain all the required information items, and the procedure is performed privately. The input voice contains the information required to specify the intellectualized data. When the news item, using the knowledge of the information item data, in accordance with the order of the programs carried out by dialogue. This aspect uses a knowledgeable database that is universal and that correctly and quickly recognizes input sounds in place of the connected database used in the first aspect above. Therefore, in this aspect, the computer uses the intellectualized data stored in the knowledge database to talk to people. To create this knowledgeable database, first, the words that appear in the data that belong to the context (topic) in which the dialogue is scheduled are classified by name. Travel-related scenarios can use travel-related reports, books, outlines of roads, brochures, and other materials. Attributes include the name as a classification name and the type, size, and scope of the name. In the above example, "destination" appears as the "proper noun" of "text" or "noun indicating place". Hawaii, Switzerland, & United States, Ruble Pavilion (Please read the notes on the back before filling out this page) -install · -ue line
• I 1 - I I •m m · 本纸張尺度適用中國國家標準(CNS ) A4規格(210X297公慶) •15- 4736?{〇 五、發明説明(丨3 ) 經濟部智慧財產局員工消費合作社印製 =外’極光可見之處、有冰河之處等。「曰數」係以數 :與曰與月結合。又’「出發曰(時)」係年、月、曰、時、 數字之組。’巾_」之前出現之數字係、對該年或表 不=年之數字加上丨者。「月」則限於數字丨至^,「日」 之前之數字祕於〗至31。該等條件可於欲進行單詞分解及 /或文章解析以正確且迅速地辨識輸入之聲音時發揮重要 的作用·。又,在上述第丨及第2樣態中所使用之關連式資料 庫則相當於在知識化資料之屬性中未記載此等條件者。因 此’實際上’可使用關連式資料庫而實行之所有操作程序, 亦可實行於知識化資料上。 一又,等級係屬性之一要素之名稱值,知識化單詞即係 藉屬性與等級之組合而構成者。CPU之控制裝置則將該等 知識化單詞儲存於記憶體而作成知識化單詞辭典。 知識化資料係使用該單詞辭典而作成,並由眾多屬性/ 等級之組合所構成,再被記錄·儲存,而構築成知識化資 料庫並記錄於記憶體中者。此等知識化資料並被授與用以 與他者識別之ID而儲存於知識化資料庫中。m可直接利用 旅行時使用於小冊子等之旅行行程代碼,或亦可利用其他 之識別號碼。在上述之旅行例中,相對於行程代碼『th A ^ 5N』之屬性之名稱「目的地」、「目的」、 發曰(時)」、「人數」、「價格」、 店」、「房間的種類」、「有無附餐」 、「曰數」、 「航空公司」、 ‘「有無自費活動 厂 . 裝 訂 —線 (請先閱讀背面之注意事項再填寫本頁) 護照 、「付費方式」,記錄有「夏威夷 「豪華旅行」、「6曰」、「2000年9月15曰上午9時00分 簽證 本紙張尺度適用中國國家標準(CNS) A4規格(210x297公釐) -16- 47367b A7 --一____B7____ 五、發明説明(l4 ) 「2以上」、「20〜25萬」、「JET或£NA」、「榭拉恩· 歡奇飯店」、「豪華等級套房」'「可選擇」、「可選擇」、 「可選擇」、「需要」、「各種信用卡或現金」。上述之 「可逞擇」者則將視使用者之回答而確定資訊項。 電腦之記憶體中並可記錄設定了以各屬性為話題時之 次序的對話順序及各屬性成為話題後之措辭等之程式。 以一上為實施本樣態前預先準備之程序。人與電腦之對 話則於該等準備完成後進行。電腦將利用上述之知識化單 詞辭典對人類使用者所輸入之聲音進行單詞分解及/或文 章解析,而判斷有無與各屬性相對應之資訊項。在使用上 述之知識化單詞辭典對輸入之聲音進行單詞分解及/或文 章解析此點上係與上述之第丨樣態不同,其他部分則與第! 樣態之發明相同。即,當輸入之聲音中未包含用以指定知 識化資料庫中之知識化資料(相當於記錄)所需之資訊項 4,就欠缺之資訊項徵詢使用者以取得該所需之全部資訊 項。另,當輸入之聲音中包含用以指定知識化資料所需之 為訊項時,使用該知識化資料之資訊項,依照對話順序使 程式進行。 在本樣態中,屬性包含有名稱與條件,因此,必須對 等級輸入符合其條件之值。而當cpu檢索知識化單詞辭典 時,由於將僅抽出相同名稱且滿足受指定之條件者,故該 檢索時間將大幅縮短,同時精確度也將大幅提昇。 本發明之申請專利範圍第7項係如申請專利範圍第6項 之主題對話方法,其中該欠缺之資訊項之徵詢程序係將屬 本紙張尺度適用中國國家標準(CNS )八视^ ( 21〇χ297公疫) B顧 ------- f請先閲讀背面之注意事項再填寫本頁) 裝· 訂 經濟部智慧財產局員工消費合作社印製 •17- 47367{。• I 1-II • mm • This paper size is applicable to Chinese National Standard (CNS) A4 specification (210X297 public holiday) System = Outside 'where Aurora is visible, where there are glaciers, etc. "Yue number" refers to number: combined with Yue and month. Also, "departure (time)" is a group of year, month, day, hour, and number. The number appearing before "Towel_" is the number of the year or the year = plus the number. "Month" is limited to the numbers 丨 to ^, and the numbers before "Day" are secreted from 〖to 31. These conditions can play an important role when you want to perform word decomposition and / or parsing to correctly and quickly recognize the input voice. In addition, the relational database used in the above-mentioned second and second aspects is equivalent to those in which the conditions are not recorded in the attribute of the knowledgeable data. Therefore, 'actually' all operational procedures implemented using connected databases can also be implemented on knowledgeable materials. Again, the name value of one element of the hierarchy attribute, the knowledge word is formed by the combination of the attribute and the hierarchy. The control device of the CPU stores the knowledgeable words in a memory to form a knowledgeable word dictionary. Knowledgeable data is created using this word dictionary, and it is composed of a combination of many attributes / levels, which is then recorded and stored, and is constructed into a knowledgeable database and recorded in memory. These knowledgeable materials are given IDs for identification with others and stored in the knowledgeable database. m can directly use the travel itinerary code used in booklets, etc., or other identification numbers. In the above travel example, the names "Destination", "Destination", "Fare (Hour)", "Number of People", "Price", "Shop", and "Room" relative to the attributes of the itinerary code "th A ^ 5N" Type "," with or without meals "," number "," airline "," "is there any self-funded activity factory. Binding-line (please read the precautions on the back before filling this page) passport," payment method ", Recorded "Hawaii" Luxury Travel "," 6th "," 9th, September 15th, 2000, visas at 9:00 am This paper size applies to the Chinese National Standard (CNS) A4 specification (210x297 mm) -16- 47367b A7- -一 ____B7____ 5. Description of the Invention (l4) "2 or more", "200,000 to 250,000", "JET or £ NA", "Xielan Hotel, Hotel", "Luxury Class Suite" "" Optional " , "Optional", "optional", "need", "various credit cards or cash". The above "optional" will determine the information items based on the user's response. The computer's memory can record programs that set the order of dialogue when each attribute is the topic and the wording after each attribute becomes the topic. The above is the procedure prepared in advance before implementing this aspect. The dialogue between the person and the computer takes place after such preparations are completed. The computer will use the above-mentioned knowledgeable word dictionary to perform word decomposition and / or text analysis on the sound input by the human user, and determine whether there is an information item corresponding to each attribute. The use of the above-mentioned knowledgeable word dictionary for word decomposition and / or text analysis of the input voice is different from the above-mentioned aspect, and the other parts are different from the first! The invention of the aspect is the same. That is, when the input voice does not contain the information items 4 required to specify the knowledgeable data (equivalent to records) in the knowledgeable database, the user is consulted for the missing information items to obtain all the required information items . In addition, when the inputted voice contains information items required to designate the knowledgeable data, the information items of the knowledgeable data are used to execute the program in accordance with the dialogue order. In this aspect, attributes include names and conditions, so you must enter a value for a level that meets its conditions. When the cpu searches for a dictionary of knowledgeable words, since only those with the same name will be extracted and meet the specified conditions, the search time will be greatly shortened, and the accuracy will be greatly improved. Item 7 of the scope of patent application of the present invention is the subject dialogue method as described in item 6 of the scope of patent application, in which the inquiry procedure of the missing information item will be in accordance with the Chinese national standard (CNS) eight views ^ (21〇 χ297 public epidemic) B Gu ------- f Please read the notes on the back before filling out this page) Binding and printing Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs • 17-47367 {.
經濟部智慧財產局員工消費合作社印製 性名稱加入徵詢句以進行詢問者。 使用者則可由徵詢句之「屬性之名稱」正確地把_ 問之重點,而確實取得意向之構通。舉例言之,可對使用 者詢問「目的地是哪裡呢?」、「曰數為幾天呢?」、「出 發曰(時)為何時呢?」。 本發明之第4樣態可提供一種主題對話方法,係用於電 腦與\間H理智並富於變化且具深度之主題對話者, 包含有:一知識化單詞辭典作成程序,係將屬特定情境(話 題)之各資料中出現之單詞,依各名稱加以分類,並以作為 該分類名之名稱與該名稱之類型、大小及可取得之範圍等 條件為其屬性,且以該名稱之值作為等級之知識化單詞加 以累積,而作成知識化單詞辭典者;一記憶體記錄程序, 係於電腦之δ己憶體中記錄具有可識別各情境(話題)之目錄 之諸多知識化資料之屬性及等級,以及,由用以規定與預 疋之属性及/或4級相關連之小情境之知識化資料之諸多 項目所構成之構造事例及其内容之内容事例,同時記錄設 定了以各屬性及項目為話題時之次序的對話順序及項目順 序’以及,各屬性及項目成為話題後之措辭等之程式者; 一詢問程序,係由電腦向人詢問所欲目錄,以指定其為何 種情境之對話者;一快取記憶體記錄程序,係依照輸入*之 目錄指定情境’再由記憶體叫出包含相同情境之相同屬性 組之多筆或單筆知識化資料而將之記錄於快取記憶體中, 且使其呈可自由改寫之狀態者;一判斷程序,係利用知識 化單詞辭典對使用者所輸入之聲音進行單詞分解及/或文 本紙張尺度適用中國國家標準(CNS ) Α4規格(210Χ297公餐) 裝------訂------線 (請先聞讀背面之注意ί項再填寫本頁} -18- 473671) 經濟部智慧財產局員工消費合作社印製 ----- 五、發明説明(π ) 讀析,以卩斷有無與各屬性相對應之資訊項纟;-徵詢 =,係於輸人之聲音中未包含用以指定知識化資料所需 '訊項時,就欠缺之資訊項徵詢使用者,以取得該所需 邦資訊項目者;一程式進行程序,係於輸入之聲音; ί含用以指定知識化資料所需之資訊項時,使用該知識化 二、;斗之 > 訊項,依照對話順序使程式進行者;一次常式進 仃%序、,係於預定之屬性及/或等級成為話題時,叫出用以 規定小情境之關連式資料庫以作為次常式,而依照項目順 序使程式進彳了者;&,—餘式進行程序,係於項目順序結 束¥回復為對話順序而使其餘程式進行者。 本發明之申請專利範圍第9項係如申請專利範圍第8項 之主題對話方法,其中該構造事例係指對話例。 本發明之申請專利範圍第1〇項係如申請專利範圍第 或第9項之主題對話方法,其中該小情境並具有居其下位 小It i見,又,該居於下位之小情境又具有更居其下位之 情境,而形成無限之階層者。 本發明之申請專利範圍第Π〜20項係有關用以實行 凊專利範圍第1〜10項之主題對話方法之主題對話系統者 本發明之第5態樣可提供一種主題對話系統,係用以 電腦與人間進行自然且理智之主題對話者,包含有··一 音輸入轉換裝置,係用以將使用者的聲音轉換成電腦之⑽ 入信號者;一聲音輸出裝置,係用以將電腦之輸出信號轉 換成聲音而輸出者;一記憶體,係用以記錄關連式資料庫 之綱目及元組,同時記錄設定了以各方案為話題時之次序 8 之 中 使 聲 I--------1衣------IT------0 (請先閲讀背面之注意事項再填寫本頁) -19· 473671) A7 " ' ----------_ 五、發明説明(π ) '— --— (請先閱讀背面之注意事項再填寫本頁) 的對話順序及各方案成為話題後之措辭等之程式者:一資 訊項判斷控制裝置,係用以對使用者輸入之聲音進行显气 分解及/或文章解析,以判斷有無與各方案相對應之資訊 項者;一必要資訊項徵詢控制裝置,係用以於輸入之聲音 中未:含用以指定關連式資料庫中之記錄所需之資訊項 時,就欠缺之資訊項徵詢使用者,以取得該所需之全部資 訊項目_者;及,-程式進行控制裝置,係用以於輸入之聲 2中包含有用以指定記錄所需之資訊項時使用該記錄之 資訊項’依照對話順序使程式進行者。 本發明之申請專利範圍第12項係如申請專利範圍第i i 項之主題對話系統,其中該必要資訊項徵詢控制裝置係可 將方案名稱加入徵詢句以進行詢問,而藉此進行徵詢欠缺 之資訊項者。 經濟部智慧財產局員工消費合作社印製 本發明之第6態樣可提供一種主題對話系統,係用以使 電腦與人間進行自然且理智之主題對話者,包含有··一聲 音輸入轉換裝置,係用以將使用者的聲音轉換成電腦之輸 入k號者;一聲音輸出裝置,係用以將電腦之輸出信號轉 換成聲音而輸出者;一記憶體,係用以記錄具有可識別各 情境(話題)之目錄之諸多知識化資料之屬性及等級,以 及,由用以規定與預定之情境之預定方案及/或元組相關^連 之小情境之關連式資料庫之諸多項目所構成之構造事例及 其内容之内容事例,同時記錄設定了以各方案及項目為話 題時之次序的對話順序及項目順序,以及,各計劃及項目 成為話題後之措辭等之程式者;一目錄詢問控制裝置,係 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) •20- 4736?{〇 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明説明(IS ) 由電腦向人詢問所欲之目錄,以指定其為何種情境之對話 者;一快取記憶體記錄控制裝置,係可依照輸入之目錄指 定情境,再由記憶體叫出被指定之關連式資料庫而將之記 錄於婢取記憶體中,且使其呈可自由改寫之狀態者;一資 訊項判斷控制裝置,係用以對使用者輸入之聲音進行單^ 分解及/或文章解析,以判斷有無與各方案相對應之資訊 項者、一必要資訊項徵詢控制裝置,係用以於輸入之聲音 中未包含用以指定關連式資料庫中之記錄所需之資訊項 時,就欠缺之資訊項徵詢使用者,以取得該所需之全部資 訊項目者,一耘式進行控制裝置,i係用以於輸入之聲音中 包含有用以指定記錄所需之資訊項時,使用該記錄之資訊 項,依照對話順序使程式進行者;_次常式進行控制裝置, 係用以於預疋方案及/或元組成為話題時,%出用以規定小 if i見之關連式貝料庫以作為次常式,而依照項目順序使程 式進行者·’及,一回復對話順序控制裝置,係用以於項目 R序、。束時’回復為對話順序而使其餘程式進行者。 本發月之申听專利範圍第14項係如中請專利範圍第U 或第14項之主題對話系統,#中該小情境並具有居其下位 之小情境,又,該居於下位之小情境又具有更居其下位之 小情境,而形成無限之階層者。 本七月之第7 g樣可提供—種主題對話系統,係用以使 電腦!亡間進行自然且理智之主題對話者,包含有:一知 識:匕早㈣典作成控制裝置,係可將出現於屬預定對話所 進仃之It k(話題)之資料中之單詞,依各名稱加以分類, 本紙張尺度適财ϋ目 210X297公釐) 裝------訂------線 (請先閱讀背面之注意事項再填寫本頁) •21· 4736?{〇 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明説明(19 並以作為該分類名之名稱與該名稱之類型、大小及可取得 之祀圍等條件為其屬性,且以該名稱之值作為等級之知識 化單调加以累積,而作成知識化單詞辭典者;-聲音輸入 轉換裝置’係用以將使用者的聲音轉換成電腦之輸入信號 者,聲音輸出裝置,係用以將電腦之輸出信號轉換成聲 音而輸出纟;—記憶體,係用以記錄包含相同屬性組之多 個^個知識化資料庫所具有之諸多屬性/等級組,同時記 錄σ又疋了以各屬性為話題之次序的對話順序及各屬性成為 話題後之措辭等之程式者資訊項判斷控制裝置,係可 利用知識化單巧辭典對使用者所輸人之聲音進行單詞分解 及/或文章解析,以判斷有無與各屬性相對應之資訊項 者;一必要資訊項徵詢裝置,係用以於輸入之聲音中未包 含用以指定知識化資料所需之資訊項時,就欠缺之資訊項 徵詢使用纟,以取得該所需之全部資訊項目者;及,一程 式進行控制裝置,係、用以於輸入之聲音中包含用以指定知 識化資枓所需之資訊項時,使用該知識化資料之資訊項, 依照對逢順序使程式進行者。 本㈣之申請專利範圍第17項係如申請專利範圍第μ 將對㈣統,其中該必要資訊項徵詢控制裝置係可 將屬性名稱加入徵詢句以進行詢問,而藉此進行徵詢欠.缺 之貧訊項者。 本發明之第8態樣可提供一種 電腦與人間進行自然且理智之主題二對 識化單詞辭典作成控制裝置,係匕3有 ★ Λ將屬特定情境(話題) — 裝 I ―― I I 訂—— I n I fe» (請先閱讀背面之注意事項再填苟本頁) -22- 經濟部智慧財產局員工消費合作社印製 473671) A7 ___ B7_ 五、發明説明(2〇 ) 之各資料中出現之單詞,依各名稱加以分類,並以作為該 为類名之名稱與該名稱之類型、大小及可取得之範圍等條 件為其屬性,且以該名稱之值作為等級之知識化單詞加以 累積*,而作成知識化單詞辭典者;一聲音輸入轉換裝置, 係用以將使用者的聲音轉換成電腦之輸入信號者;一聲音 輸出裝置’係用以將電腦之輸出信號轉換成聲音而輸出 者;一-記憶體,係用以記錄具有可識別各情境(話題)之目 錄之諸多知識化資料之屬性及等級,以及,由用以規定與 預疋之屬性及/或專級相關連之小情境之知識化資料之諸 多項目所構成之構造事例及其内容I之内容事例,同時記錄 設定了以各屬性及項目為話題時之次序的對話順序及項目 順序,以及,各屬性及項目成為話題後之措辭等之程式者; 一目錄詢問控制裝置,係由電腦向人詢問所欲之目錄,以 才曰疋其為何種情境之對話者;一快取記憶體記錄控制裝 置’係可依照輸入之目錄指定情境,再由記憶體叫出包含 相同屬性組之多筆或單筆知識化資料而將之記錄於快取記 憶體中,且使其呈可自由改寫之狀態者;一資訊項判斷控 制裝置,係可利用知識化單詞辭典對使用者所輸入之聲音 進行單詞分解及/或文章解析,以判斷有無與各屬性相對 應之資訊項者,一必要資訊項徵詢裝置,係用以於輸入之 聲音中未包含用以指定知識化資料所需之資訊項時,就欠 缺之資訊項徵詢使用者,以取得該所需之全部資訊項目 者;一程式進行控制裝置,係用以於輸入之聲音中包含用 以指定知識化資料所需之資訊項時,使用該知識化資料之 裝 訂 線 f請先閱讀背面之注意事項再填寫本頁)The printed consumer's name of the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs was added to the inquiry sentence for inquiry. The user can use the "name of the attribute" in the query to correctly focus on the question, and indeed obtain the intention. For example, you can ask the user "Where is the destination?", "How many days is it?", "When is the date of delivery?". The fourth aspect of the present invention can provide a topical dialogue method, which is used for computer and interpersonal intelligent and changeful and in-depth topical interlocutors, including: a knowledgeable word dictionary creation process, which will be specific The words appearing in each data of the context (topic) are classified according to each name, and the attributes such as the name of the classification name and the type, size, and available range of the name are its attributes, and the value of the name is used Those who accumulate knowledgeable words as grades and create a dictionary of knowledgeable words; a memory recording program, which is a computer-recorded δ-memory body that records many attributes of knowledgeable data with a catalog that can identify each situation (topic) And grades, as well as content cases that consist of a number of items that specify knowledgeable data related to the attributes and / or level 4 of the forecast, as well as content examples of the content, while recording the setting of each attribute And the order of the conversation when the item is the topic and the order of the item 'and the formulator of the wording after each attribute and item becomes the topic; The computer asks the person for the desired directory to specify the contextual interlocutor; a cache memory recording program is to specify the context according to the directory input *, and then the memory calls out the same attribute group containing the same attribute group of the same context. Those who record multiple or single pieces of intellectualized data in cache memory and make them freely rewriteable; a judgment procedure that uses the knowledgeable word dictionary to decompose words input by users And / or text paper size applicable to China National Standard (CNS) Α4 specification (210 × 297 catering) Packing ------ order ------ line (please read the note on the back before filling this page} -18- 473671) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs ----- 5. Read and analyze the invention description (π) to determine whether there are information items corresponding to each attribute;-Consultation =, related to When the input voice does not include the 'information' required to specify the intellectualized data, the user is consulted for the lack of information items to obtain the required information item; a program performs the procedure, which is based on the input voice; ί Contains the designated knowledge resources When the required information item is used, use the knowledgeable two; Dou > message item to make the program performer in accordance with the dialogue order; once a routine enters the% order, when a predetermined attribute and / or level becomes a topic, Relevant database for defining small situations is used as a subroutine, and the program is entered in accordance with the order of items; &, — The remainder process is performed at the end of the item order. The rest of the program is performed. Item 9 of the scope of patent application of the present invention is the subject dialogue method as described in item 8 of the scope of patent application, wherein the construction example refers to a dialogue example. Item 10 of the scope of patent application of the present invention is the method of topical dialogue as described in the scope of application or item 9 of the patent scope, in which the small situation has a lower ranking, and the small situation Those who are in a subordinate situation and form an infinite class. The present invention has a range of Π to 20 in the scope of the patent application of the present invention. The subject dialogue system is used to implement the theme dialogue method of 凊 patents 1 to 10. The fifth aspect of the present invention can provide a theme dialogue system for A natural and sane subject interlocutor between a computer and a human, including a sound input conversion device for converting the user's voice into a computer's input signal; a sound output device for converting the computer's The output signal is converted into sound and output; a memory is used to record the outlines and tuples of the relational database, and at the same time records the order when the various topics are set. --- 1 clothing ------ IT ------ 0 (Please read the precautions on the back before filling this page) -19 · 473671) A7 " '--------- -_ V. Description of the invention (π) '— --- (Please read the notes on the back before filling out this page) Dialogue order and the formulas for the wording after each plan becomes the topic: an information item judgment control device, It is used to decompose the sound input by the user and / or analyze the article to determine whether there is There is no information item corresponding to each scheme; a necessary information item consultation control device is used in the input voice: when it contains the information items required to specify the records in the connected database, it is lacking. The information item consults the user to obtain all the required information items_; and-the program control device is used to input the recorded sound 2 when it contains useful information items required to specify the record The information item 'plays the programmer in the order of the conversation. Item 12 of the scope of patent application of the present invention is a subject dialogue system such as item ii of the scope of patent application, wherein the necessary information item consultation control device can add a scheme name to the inquiry sentence to make an inquiry, and thereby perform lack of inquiry Entrants. The sixth aspect of the invention printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs can provide a thematic dialogue system, which is used to enable computers and people to conduct natural and rational thematic dialogues, including a voice input conversion device, It is used to convert the user's voice into the computer's input k number; a sound output device is used to convert the computer's output signal into sound and output; a memory is used to record the identifiable situations Attributes and levels of many knowledgeable materials in the (topic) directory, and a number of items in a relational database that defines small contexts associated with predetermined scenarios and / or tuples of predetermined contexts Cases of construction and content, and record the dialogue order and project order that set the order of each plan and project as the topic, as well as the formulator of the wording after each plan and project becomes the topic; a directory inquiry control Device, this paper size applies Chinese National Standard (CNS) A4 specification (210X297 mm) • 20- 4736? {〇A7 B7 Ministry of Economic Affairs wisdom Printed by the Production Cooperative Consumer Cooperatives V. Invention Description (IS) The computer asks people for the desired directory to specify the situation's interlocutor; a cache memory record control device can be specified according to the entered directory The situation, and then the designated connected database is called from the memory and recorded in the capture memory, and made it in a state that can be freely rewritten; an information item judgment control device is used for the user The input sound is decomposed and / or parsed to determine whether there is an information item corresponding to each scheme, and a necessary information item consultation control device is used to specify that the input sound does not include related data. When the required information items are recorded in the library, the user is consulted for the missing information items to obtain all the required information items. The control device is a one-click type, i is used to include useful information in the input sound. When specifying the information items required for recording, use the recorded information items to make the program performers in accordance with the dialogue order; _ subroutine control device, used to predict the plan and / or When the composition is a topic, the related shellfish library used to specify the small if i sees is used as a subroutine, and the program is performed according to the order of the items. And, a reply dialog sequence control device is used for the item R order ,.时时 ’reverts to the dialogue sequence and causes the rest of the program to proceed. Item 14 of the scope of the application for the patent this month is the subject dialogue system of item No. U or No. 14 in the patent application. The small situation in # has the small situation in the lower position, and the small situation in the lower position. Those who have a lower situation, and form an infinite stratum. The 7th sample of this July can provide a kind of thematic dialogue system for the computer! The natural and rational thematic interlocutors among the dead include: a knowledge: a control device created by the dagger as early as possible, the words appearing in the data of It k (topic) entered into the predetermined dialogue, according to each Names are categorized, this paper is suitable for financial purposes, 210X297 mm) Packing ------ Order ------ Line (Please read the precautions on the back before filling this page) • 21 · 4736? {〇 A7 B7 Printed by the Consumers' Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. 5. Description of the invention (19 and the conditions as the name of the classification and the type, size, and sacrifice of the name can be used as its attributes, and Values are accumulated as knowledge monotonicity of grades to make a knowledgeable word dictionary;-sound input conversion device 'is used to convert the user's voice into a computer input signal, sound output device is used to convert the computer The output signal is converted into sound and output; 纟 memory, which is used to record many attribute / level groups of multiple ^ knowledgeable databases containing the same attribute group, while recording σ and using each attribute as Order of topics The programmer's information item judgment control device of the dialogue order and the wording of each attribute after the topic becomes the topic, can use the knowledgeable dictionary to perform word decomposition and / or article analysis on the voice input by the user to determine whether the Information items corresponding to each attribute; a necessary information item inquiry device is used to inquire for the lack of information items when the inputted voice does not contain the information items required to specify the knowledgeable data to obtain All the required information items; and a program control device for using the information items of the knowledgeable data when the inputted voice contains the information items required to specify the knowledgeable assets, The program is performed in accordance with the order of confrontation. The 17th scope of the patent application scope of this document will be the system if the patent scope scope of the patent application is μ, in which the necessary information item inquiry control device may add the attribute name to the inquiry sentence for inquiry, In this way, people who are lacking in information are consulted. The eighth aspect of the present invention can provide a natural and rational theme between the computer and the human being. The dictionary is made into a control device. It has 3 ★ ★ Λ will belong to a specific situation (topic) — Install I —— II Order —— I n I fe »(Please read the notes on the back before filling this page) -22- Economy Printed by the Ministry of Intellectual Property Bureau's Consumer Cooperatives 473671) A7 ___ B7_ V. The words appearing in the information of the invention description (20) are classified according to each name, and the name as the class name and the type of the name Conditions such as the size, size, and available range are its attributes, and the knowledgeable words with the value of the name as the level are accumulated *, and a knowledgeable word dictionary is created; a voice input conversion device is used to convert the user ’s A person who converts sound into computer input signals; a sound output device is used to convert computer output signals into sound and output; a memory is used to record a large number of directories with recognizable situations (topics) Attributes and levels of knowledgeable data, and examples of constructions made up of many items of knowledgeable data that specify small contexts related to the attributes and / or levels of advancement The content example of I also records the dialogue order and item order that set the order in which each attribute and item is the topic, and the formulae of the wording after each attribute and item becomes the topic; a catalog inquiry control device, The computer asks the person for the desired directory, and talks about what kind of situation it is; a cache memory record control device can specify the situation according to the input directory, and then the memory calls out the memory containing the same attribute group. Multiple or single pieces of intellectualized data are recorded in cache memory and made freely rewritable; an information item judgment control device can use a knowledgeable word dictionary to input information entered by users Voice decomposition and / or article analysis to determine whether there are information items corresponding to each attribute, a necessary information item consultation device, the input sound does not contain the information required to specify knowledgeable data Items, consult users for missing information items to obtain all the required information items; a program control device for inputting Sound contains information used to designate items of knowledge required information, the use of knowledge-based information stapling line f Please read the Notes on the back to fill out this page)
47367b 五、發明説明( 資訊項,依照對話順序使程式進行者;…_ 裝置,係用以於預定之屬性及/或等級成為話料進== 以規定小情境之知識化資料以作為次常 J出用 序使程式進行者;…回復對話順序控制袭置:::: 於項目順序結束時,回復為對話順序而使其餘程式進行者 本發明之申請專利範圍第19項係如申請專利範圍_ 項之主·題對話系統’其中該構造事例係指對話例。 8 本發明之申請專利範圍第2 0項係如申請專利範圍第卜 或第19項之主題對話系統,#中該小情境並具有居其下位 之小情境,又’該居於下位之小情境又具有更居其下位之 小情境’而形成無限之階層者。 以下將根據圖示之較佳實施例就本發明加以詳細說 明,圖不者僅單純為一例,而非用以限定本發明之範圍者。 應留意本發明可進行各種修正·改變而不致偏離所附之申 請專利範圍中所記載之發明精神。 【圖式之簡單說明】 經濟部智慧財產局員工消費合作社印製 第1圖係本發明之第一樣態中之主題對話方法之一實 施例之流程圖。 第2圖係顯示用以實施第1圖所示之主題對話方法之^ 統之一實施例之方塊圖。 第3圖係本發明之第2樣態中之主題對話方法之一實施 例之流程圖。 第4圖係顯示電腦對人進行旅行介紹時之對話之一例 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公餐 -24· 47367 丨j A7 B7 五、發明説明(22 者 經濟部智慧財產局員工消費合作社印製 第5圖係顯示用以實施第3圖所示之主題對話方法之系 統中主電腦9之一實施例之方塊圖。 弟6圖係本發明之第3樣態中之主題對話方法之一實施 例之流程圖。 第7圖係顯示用以實施第6圖所示之主題對話方法之系 統中主、電腦9之一實施例之方塊圖。 第8圖係本發明之第4樣態中之主題對話方法之一實施 例之流程圖。 第9圖係顯示用以實施第§圖所示之主題對話方法之 統中主電腦9之一實施例之方塊圖。 第10圖係本發明之第1樣態中具特有構造之資料庫 構築方法之一實施例之流程圖。 第11圖係顯示知識化單詞之數例之表格。 第12(a)及(b)圖係分別顯示已將舉出作為例文之文 單詞分解後之狀態之說明圖。 第13(a)及(b)圖係顯不由第12(a)及(b)圖之文書所得 知識化資料之一例之說明圖。 第14圖係顯示以往之關連式資料庫之表格。 第1 5圖係由第14圖所示之關連式資料庫選出一筆記錄 而得到之關連式資料庫之說明圖。 第16圖係由進行以第15圖中之關連式資料庫所示之一 筆或數筆記錄之元組為等級,而以綱目為屬性之名稱的旨 意置換而得到之知識化資料之說明圖。 系 之 章 之 (锖先閱讀背面之注意事項再填寫本頁) -裝.47367b V. Description of the invention (information item, the program is performed in accordance with the dialogue order; ..._ device, which is used to become a source of information at a predetermined attribute and / or level == the knowledge of the specified small context as a secondary Use the sequence to make the program performer; ... Resume the dialog sequence control and reset ::: At the end of the project sequence, return to the dialog sequence and make the rest of the program performers. _ Topic dialogue system of item 'where the construction example refers to a dialog example. 8 The scope of application for the patent of the present invention is No. 20, such as the subject dialogue system of scope No. 19 or No. 19 of the application for patent scope, # 中 此 小 Context Those who have a small situation that is inferior to them, and that "the small situation that is inferior to have a small situation that is more inferior to" form an infinite hierarchy. The invention will be described in detail below based on the illustrated preferred embodiment The diagram is only an example, and is not intended to limit the scope of the invention. It should be noted that the invention can be modified and changed without departing from the scope of the appended patent application. The spirit of the invention is contained in the invention. [Simplified description of the drawing] The first picture printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs is a flowchart of one embodiment of the subject dialogue method in the first state of the present invention. A block diagram showing one embodiment of a system for implementing the theme dialog method shown in FIG. 1. FIG. 3 is a flowchart of one embodiment of the theme dialog method in the second aspect of the present invention. The picture shows an example of a dialogue when a computer introduces people to travel. This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 public meals-24 · 47367 丨 j A7 B7. V. Description of invention (22 Intellectual Property Bureau, Ministry of Economic Affairs) Figure 5 printed by the employee consumer cooperative is a block diagram showing one embodiment of the host computer 9 in the system for implementing the theme dialogue method shown in Figure 3. Figure 6 is a theme in the third aspect of the present invention. A flowchart of one embodiment of a dialogue method. Figure 7 is a block diagram showing one embodiment of a host and a computer 9 in a system for implementing the theme dialogue method shown in Figure 6. Figure 8 is the first embodiment of the present invention. Thematic dialogue in 4 aspects A flowchart of one embodiment of the method. FIG. 9 is a block diagram showing an embodiment of the host computer 9 in the system for implementing the theme dialogue method shown in FIG. §. FIG. 10 is the first example of the present invention. A flowchart of an embodiment of a method for constructing a database with a unique structure in the state. Figure 11 is a table showing examples of knowledgeable words. Figures 12 (a) and (b) show that each has been cited as Figure 13 (a) and (b) is an explanatory diagram showing the state of the word after the text is decomposed. Figure 13 (a) and (b) are explanatory diagrams showing an example of the intellectualized data obtained from the documents of Figures 12 (a) and (b). Figure 14 It is a table showing the past related database. Figure 15 is an explanatory diagram of the related database obtained by selecting a record from the related database shown in Figure 14. Fig. 16 is an explanatory diagram of the knowledgeable data obtained by performing the intentional replacement of the tuple of one or several records shown in the relational database of Fig. 15 with the outline as the attribute name. Chapter of the system (read the precautions on the back before filling out this page)-installed.
、1T 線 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) -25- 4736?ϋ A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明説明(23 ) 第1 7圖係顯示以往之物件導向資料庫者。 第1 8圖係顯示以由共通部分所構成之部分技術化資料 為上位而以包含不同要素單元之技術化資料為下位之物件 導向資料庫之作成步驟之說明圖。 第19圖係用以說明就某單詞取得高相關度之相關語之 一方法例之表格。 第_20圖係顯示本發明中具特有構造之資料庫之構築方 法之一實施形態流程之流程圖。 第21 (a)及(b)圖係分別用以說明以往之物件導向資料 庫中之資料構造及資料間之樹狀構造之概略圖。 【本發明之較佳實施形態】 以下參照附圖,就本發明中之主題對話方法及系統之 一實施例加以詳細說明。首先,第1圖係本發明之第1樣態 中之主題對話方法之一實施例之流程圖,第2圖係顯示用以 實施第1圖所示之主題對話方法之系統之一實施例之方塊 圖。 本發明之第1樣態中之主題對話方法大致包含有資料 庫及程式之記錄程序(步驟1對輸入之聲音進行單詞分解 及/或文章解析後之資訊項有無判斷程序(步驟2);當輸1入 之聲音中是否包含用以指定記錄所需之資訊項之判斷j丨結 果為「否」時之必要資訊項之徵詢程序(步驟3);及,當判 斷J1結果為「是」時,或藉徵詢程序(步驟3)齊備用以指定 記錄所需之資訊項後之程式進行程序(步驟4)。 --------裝-- (請先閱讀背面之注意事項再填寫本頁) 訂 線 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) •26- 47367^0 五、 發明説明(24 、,在本樣態中,電腦係使用記憶體中所記錄之關連式資 料庫中所儲存之資料以與人對話。第14圖係顯示以往之典 型關連式資料庫之資料構造之表。表中,檢索鍵 之屬焯(即:方案),TU〜Tmn係其内容或值之元組。各行 皆構成一項記錄。當該關連式資料庫係有關旅行者時,該 =案S1〜Sn即為諸如「目的地」、「目的」、「曰數」、 出發·日(時)」、「人數」、「價格」、「航空公司」、 「「飯店」、「房間之種類」、「有無附餐」、「自費活動」、 「護照」、「簽證」、「付費方式」等。而,各筆記錄中 則記載有與該等綱目相對應之元組、即,在各筆記錄中 可 經濟部智慧財產局員工消費合作社印製 記錄諸如「夏威夷」、「泰國普吉島」、「芬蘭赫爾辛基 「京都」、「陸奥之青森」、「沖繩」、「英國」等目 地各異之旅行資訊。 又,記憶體中亦記錄有已設定與人對話時以何順序將 各方案加人話題之對話順序及用以規定各方案成為話題後 之措辭及其變形等之程式。電腦之cpu則可由記憶體中 出該程式而依該程式繼續進行對話。在旅行之例子中, 話順序即以對人詢問「請問您需要什麼服務?」之會話句 開始進行對話。 在資訊項有無判斷程序(步驟2)中,一旦人對電腦發 聲音,電腦即運用麥克風、聲音辨識軟體等以辨識音素 舉例言之,如「響去下威宜」、「響看急光」#。再利,, 單詞辭典、構句辭典、事例辭典等而對其進行單詞分解及 或文章解析以正確辨識成「想去夏威夷」、「想看極光 的 叫 對 而 出 用 ---------1------IT------# (請先閲讀背面之注意事項再填寫本頁) -27· 經濟部智慧財產局員工消費合作社印製 47367{〇 五、發明説明(25) 「接者,以判_判斷與各方案相對應之值(此處稱為 資訊項」)是否存在於業經單詞分解及/或文章解析之輸 入之聲音中。@ ’為便於決^輸人之聲音所要求者為關連 式資料庫中之何記錄,在關捸咨处 τ社關運式貝枓庫中儲存有眾多記 、彔即w輸入之聲音中未包含用以指定記錄所需之資訊 項時’即可藉徵詢程序(步驟3)就欠缺之資訊項徵詢使用者 以取得-該所需之全部資訊項。藉此,即可指;ti筆(或數筆) 記錄。 舉例《之,有關旅行之關連式資料庫中储存有數十〜 數百之以夏威夷為目的地之旅行資訊。欲由其中選定使用 者所欲之記錄需要更進一步之資訊項。即「目的」、「曰 數」、「出發日」等。而,程式中已事先決定何等方案之 元組將成為此等必要資訊項,而可判斷業經單詞分解及/ 或文章解析之輸入之聲音中是否自&即全部齊冑帛以指定 此專記錄之資訊項。 當用以指定記錄所需之資訊項自始即包含於輸入之聲 曰中日守,或,以上述程序徵詢使用者而齊備之後,即在程 式進行程序(步驟4)中使用該記錄之資訊項,電腦之CPU則 依照對話順序而使程式進行。通常,在對話順序中係使用 與該ό己錄之全部或適合之方案相對應之資訊項而依照預^定 之順序使對話進行。 在圖示之較佳實施例中,欠缺之資訊項之徵詢程序(步 驟3)係將方案名稱加入徵詢句以進行詢問者。而,上述「想 去夏威夷」之使用者的回答並未包含「目的」、「日數」、 本紙張尺度適用中國國家標準(CNS ) Α4規格(210X29*7公釐) -----------^------1Τ------# (請先閱讀背面之注意事項再填寫本頁) -28· 經濟部智慧財產局員工消費合作社印製 473671) A7 ____—_B7_ 五、發明説明(26 ) 「出發曰」等必要資訊項。因此,電腦將徵詢該等資訊項, 此時,宜如「您(旅行)的目的為何?」、「日數為幾天?」、 出發日為何時?」般,將方案名稱直接加入徵詢句以進行 詢問Γ藉此,可避免電腦與人間之對話中產生意向溝通不 良之問題。進而,亦可如「旅行之目的請由蜜月旅行、海 上活動、日光浴、購物、洽商、其他等選擇其一。」般, 限定使、用者之回答以提昇辨識精確度。 以下參照第2圖,其係顯示用以實施第1圖所示之主題 對話方法之系統之一實施例者。 另’在圖示之較佳實施例中,雖顯示了藉網路而實施 主題對話方法之系統之一實施例,但理當不限於此。舉例 言之,使電腦進行公司之詢問業務之使用形態則未透過網 路,而由人與電腦直接進行對話。 本發明之主題對話系統係大致包含麥克風等聲音輸入 裝置1 ’喇队、耳機等聲音輸出裝置3,用戶終端5,網際網 路、企業網路、LAN等通訊電路7及用以管理本系統之主電 腦9 〇 聲音輸入裝置1係用以將使用者人類所發出之聲音轉 換成可藉電腦加以處理之數位信號者。聲音輸出裝置3係可 依據由電腦生成之聲音發生信號而將數位信號轉換成聲音 者。用戶終端5係可藉迄今為人所熟知之各種個人電腦與網 際網路相連接者。用戶終端5之處理結果將藉通訊電路傳輸 至主電腦9,而,主電腦9之處理結果則將藉通訊電路7為用 戶終端5所接收。 本紙張尺度適用中國國家標準(CNS ) A4規格(21 Οχ 297公楚) ---------^------1Τ------^ (請先閱讀背面之注意事項再填寫本頁) -29· 4736?{〇 經濟部智慧財產局員工消費合作社印製 A7 B7 五、發明説明(27 ) 主電腦9内設有用以記錄各種資料及程式之記憶體i j 與可叫出記憶體中所記錄之程式以進行各種控制之 CPU 13。記憶體11中則設有由眾多方案所構成,即,由綱 目及元組所構成之關連式資料庫1丨a與用以記錄已設定以 何順序將各方案加入話題之程式之對話順序部nb,以及用 以記錄已設定各方案成為話題後之措辭等之程式之措辭記 錄部11 c。 CPU13内則設有一資訊項判斷控制裝置13a,係用以對 使用者輸入之聲音進行單詞分解及/或文章解析,以判斷 有無與各方案相對應之資訊項者;一必要資訊項徵詢控制 裝置13b’係用以於輸入之聲音中未包含用以指定關連式資 料庫中之記錄所需之資訊項時,就欠缺之資訊項徵詢使用 者’以取得該所需之全部資訊項目者;及,一程式進行控 制裝置13c,係用以於輸入之聲音中包含有用以指定記錄所 需之資訊項時,使用該記錄之資訊項,依照對話順序使程 式進行者。 在圖示之較佳實施例中,必要資訊項徵詢控制裝置! 3b 係可將方案名稱加入徵詢句以進行詢問,而藉此進行徵詢 欠缺之資訊項者。 第3圖係本發明之第2樣態中之主題對話方法之一實*施 例之流程圖。 竿5圖係顯示用以實施第3圖所示之主題對話方法之系 統中主電腦9之一實施例之方塊圖。 本發明之第2樣態中之主題對話方法係用於電腦與人 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) 裝------訂------線 (請先閲讀背面之注意事項再填寫本頁) •30· 4736?lo 經濟部智慧財產局員工消費合作社印製 A7 B7 五、發明説明(28 ) 間自然且理智之主題對話者,而與第1樣態之主題對話方法 相較,其不同之處在於電腦可由人所發出之聲音而指定情 境(話題)以進行對話,及,可於進行有關特定情境之對話 時插入可更為深入該話題之主題性小對話。 本樣態中之主題對話方法大致包含有:一資料庫及程 式之記錄程序(步驟11),一用以指定情境之目錄詢問程序 (步驟1〜2),一將受指定之關連式資料庫記錄於快取記憶體 之记錄程序(步驟13),一對使用者所輸入之聲音進行單詞 分解及/或文章解析後之資訊項有無判斷程序(步驟14),一 當輸入之聲音中是否包含用以指定:記錄所需之資訊項之判 斷jii結果為「否」時之必要資訊項之徵詢程序(步驟15), 一當判斷J11結果為「是」時,或藉徵詢程序(步驟15)而齊 備用以指定記錄所需之資訊項後之程式進行程序(步驟 16),一於預定之方案及/或元組成為話題後,轉換成有關 小情i兄之次常式之轉換程序(步驟丨7)及一於小情境之次常 式結束後,回復為原本之對話順序而使其餘程式進行之餘 式進行程序(步驟18)。 在本樣態之記錄程序(步驟n)中,電腦之記憶體中可 兄錄儲存眾多關連式資料庫,並對該等資料庫分別授與可 識別其他資料庫與該資料庫中所處理之情境(話題)之目 錄。電腦之記憶體中並記錄有用以規定與預定之方案及/ 或元組相關之小情境之關連式資料庫。用以規定小情境之 關連式資料庫則係由諸多項目所構成之構造事例(相當於 綱目)及其等之内容之内容事例(相當於元組)所構成者 ---------^------,玎------# ί請先閲讀背面之注意事項再填寫本頁}1. The paper size of the 1T line is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) -25- 4736? 7 A7 B7 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (23) Figure 17 Show past object-oriented databases. Fig. 18 is an explanatory diagram showing the steps of creating a database with a part of the technical data composed of common parts as the upper level and a technical data including different element units as the lower level. Fig. 19 is a table illustrating an example of a method for obtaining a high-relevance related phrase for a word. Fig. _20 is a flow chart showing an embodiment of a method for constructing a database with a unique structure in the present invention. Figures 21 (a) and (b) are schematic diagrams for explaining the data structure in the conventional object-oriented database and the tree structure between the data, respectively. [Preferred embodiment of the present invention] Hereinafter, an embodiment of the subject dialogue method and system in the present invention will be described in detail with reference to the drawings. First, FIG. 1 is a flowchart of an embodiment of the theme dialog method in the first aspect of the present invention, and FIG. 2 is a diagram showing an embodiment of a system for implementing the theme dialog method shown in FIG. 1. Block diagram. The subject dialogue method in the first aspect of the present invention generally includes a database and a program for recording a program (step 1 performs a word decomposition on the input voice and / or a program for judging whether an information item is parsed after the article (step 2); Enter whether the input voice contains a judgment procedure for specifying the information items required for recording j 丨 the necessary information item consultation procedure when the result is "No" (step 3); and, when the judgment J1 result is "Yes" , Or borrow the consultation procedure (step 3) to prepare for the procedure after specifying the information items required for recording (step 4). -------- install-(Please read the precautions on the back before filling (This page) The paper size of the booklet is in accordance with the Chinese National Standard (CNS) A4 (210X297 mm). • 26- 47367 ^ 0 V. Description of the invention (24. In this case, the computer is recorded in the memory. The data stored in the related database is used to talk to people. Figure 14 is a table showing the data structure of a typical connection database in the past. In the table, the search key belongs to 焯 (that is, the scheme), TU ~ Tmn A tuple of its contents or values. Each line constitutes a Records. When the related database is related to travelers, the cases S1 ~ Sn are such as "destination", "purpose", "number", departure · day (hour), "number of people", " "Price", "Airline", "Hotel", "Type of room", "With or without meals", "Self-funded activities", "Passport", "Visa", "Payment method", etc., and in each record Then records the tuples corresponding to these outlines, that is, in each record, printed records such as "Hawaiian", "Phuket, Thailand", "Kyoto, Helsinki, Finland, Travel information of various destinations such as "Atsunomori Aomori", "Okinawa", "United Kingdom", etc. In addition, the memory also records the order in which conversations are added to the topics and the order in which they are added to the conversation. Programs that formulate the wording after the topic becomes a topic, its variants, etc. The computer CPU can get the program out of memory and continue the conversation according to the program. In the travel example, the order of words is to ask people " Ask what service do you need? ”To start the conversation. In the information item judgment process (step 2), once a person makes a sound on the computer, the computer uses a microphone, voice recognition software, etc. to identify the phonemes as an example, such as 「响 去 下 威 宜」, 「响 看 急 光」 #. Zaili, word dictionary, sentence dictionary, case dictionary, etc., to perform word decomposition and / or article analysis to correctly identify as "want to go to Hawaii", "If you want to see the aurora's righteousness and use it --------- 1 ------ IT ------ # (Please read the precautions on the back before filling out this page) -27 · Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 47367 {05. Invention Description (25) "Receiver, to judge_determine whether the value corresponding to each scheme (herein referred to as the information item) exists in the industry word decomposition And / or the sound of the parsed input. @ 'In order to facilitate the determination of the input voice, what is required is a record in the relational database. There are many records stored in the Guanyun Consultation Office τ society's transport-type shell library, that is, the voice input by w When it contains the information items required to specify the record, the user can be consulted for the missing information items through the consultation process (step 3) to obtain all the information items required. With this, you can refer to; ti pen (or several) records. For example, "A travel-related database stores dozens to hundreds of travel information destined for Hawaii. Further records are needed for records that are intended to be selected by the user. It means "purpose", "number", "departure date", and so on. However, the tuples in the program that have previously decided which schemes will become such necessary information items, and can judge whether the input voices that have undergone word decomposition and / or article parsing are all & that is, all of them are specified to specify this special record Information item. When the information items required to specify the record are included in the input voice from the beginning, or after the user is prepared by consulting the above procedure, the recorded information is used in the program execution procedure (step 4). Item, the computer's CPU executes the program in accordance with the dialogue order. Generally, the dialogue order is carried out in a predetermined order by using information items corresponding to all or suitable schemes of the record. In the illustrated preferred embodiment, the missing information item inquiry procedure (step 3) is to add the scheme name to the inquiry sentence for the inquiry. However, the answers of the users who want to go to Hawaii did not include "purpose", "days", and this paper size applies the Chinese National Standard (CNS) Α4 specification (210X29 * 7 mm) ------ ----- ^ ------ 1Τ ------ # (Please read the notes on the back before filling out this page) -28 · Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 473671) A7 ____ —_B7_ V. Explanation of the Invention (26) The necessary information items such as "Departure". Therefore, the computer will ask for such information items. At this time, it should be like "What is the purpose of your (travel)?", "How many days are you?", When is the departure day? In general, the name of the plan is directly added to the inquiry sentence for questioning. This can avoid the problem of poor communication in the dialogue between the computer and the person. Furthermore, you can also choose “honeymoon trips, maritime activities, sunbathing, shopping, business negotiation, etc.” to limit the answers of users and users to improve the accuracy of identification. Reference is now made to Fig. 2 which shows an embodiment of a system for implementing the subject dialogue method shown in Fig. 1. In addition, in the illustrated preferred embodiment, although an embodiment of a system for implementing a topic dialogue method via a network is shown, it is not limited thereto. For example, the use of computers to conduct the company's inquiry business is not through the network, and people and computers directly talk. The subject dialog system of the present invention generally includes a sound input device 1 such as a microphone, a sound output device 3 such as a squadron, a headset, a user terminal 5, an Internet, an enterprise network, and a communication circuit 7 such as a LAN, and a system for managing the system. The host computer 90 sound input device 1 is used to convert the sound emitted by the user's human body into a digital signal that can be processed by a computer. The sound output device 3 can convert a digital signal into a sound based on a sound generating signal generated by a computer. The user terminal 5 can be connected to the Internet by various personal computers which are known so far. The processing result of the user terminal 5 will be transmitted to the host computer 9 through the communication circuit, and the processing result of the host computer 9 will be received by the user terminal 5 through the communication circuit 7. This paper size applies to China National Standard (CNS) A4 specification (21 〇χ 297 公 楚) --------- ^ ------ 1T ------ ^ (Please read the note on the back first Please fill in this page again for matters) -29 · 4736? {〇 Printed by A7 B7, Employee Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs V. Invention Description (27) The main computer 9 is provided with a memory ij for recording various data and programs. The CPU 13 which recalls programs recorded in the memory for various controls. The memory 11 is provided with a plurality of schemes, that is, a relational database 1 composed of outlines and tuples, and a dialogue sequence unit for recording a program in which order each scheme is added to the topic. nb, and a wording recorder 11c that records the formulas that have been used to set the wording of topics after they become topics. The CPU 13 is provided with an information item judgment control device 13a, which is used to perform word decomposition and / or article analysis on the voice input by the user to determine whether there is an information item corresponding to each scheme; a necessary information item consultation control device 13b 'is used to consult the user for missing information items when the inputted voice does not contain the information items required to specify the records in the connected database; and to obtain all the required information items; and A program execution control device 13c is used to make the program performer in accordance with the dialogue order by using the recorded information items when the inputted sound contains information items useful for specifying the records. In the illustrated preferred embodiment, the necessary information items are consulted to the control device! 3b refers to those who can add the name of the scheme to the inquiry sentence for inquiry, so as to consult the missing information items. Fig. 3 is a flowchart of one embodiment of the subject dialogue method in the second aspect of the present invention. FIG. 5 is a block diagram showing an embodiment of a host computer 9 in the system for implementing the subject dialogue method shown in FIG. The subject dialogue method in the second aspect of the present invention is used for computers and human-made paper. The standard of China National Standards (CNS) A4 (210X297 mm) is applied. (Please read the precautions on the back before filling out this page) • 30 · 4736? Lo Printed by the Consumers ’Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 V. Invention Description (28) A natural and rational interlocutor on the subject, and 1 The mode of theme dialogue is different. The difference is that the computer can specify the situation (topic) for dialogue by the voice of a person, and it can be inserted in the dialogue about a specific situation to deepen the topic. Thematic small dialogue. The theme dialogue method in this aspect roughly includes: a database and program recording procedure (step 11), a directory inquiry procedure for specifying a scenario (steps 1 to 2), and a related relational database to be designated The recording procedure recorded in the cache memory (step 13), whether the word input and / or article parsing of the information items input by the user is judged (step 14), and whether the input voice is Contains the inquiry procedure for specifying the necessary information items when the judged jii result of the required information item is recorded as "No" (step 15), when the judged J11 result is "yes", or by the inquiry procedure (step 15) ) And prepare the procedure using the program after the specified information items required for recording (step 16). Once the predetermined scheme and / or the composition of the topic is the topic, it will be converted into a subroutine of the minor love i. (Step 丨 7) and one after the end of the subroutine in the small situation, return to the original dialogue sequence and make the rest of the formulas proceed (Step 18). In this aspect of the recording procedure (step n), a number of related databases can be stored in the memory of the computer, and these databases can be separately identified to identify other databases and the data processed in the database. Directory of contexts (topics). The computer's memory also contains a relational database that is used to define small contexts related to predetermined schemes and / or tuples. The related database used to specify small situations is composed of a structural example (equivalent to the outline) composed of many items and a content example (equivalent to a tuple) of the content -------- -^ ------ , 玎 ------ # ίPlease read the notes on the back before filling in this page}
-31· 47367b A7 五 經濟部智慧財產局員工消費合作社印製 B7 發明説明(29 腦之記憶體中更記錄有已設定以各方案及項目為話題之二欠 序之對話順序及項目順序,以及,各方案及項目成為話題 後之措辭等之程式。 第4圖係顯示電腦對人進行旅行介紹時之對話之—例 者。 在目錄詢問程序(步驟12)中,首先,進行由電腦向人 詢問所_欲目錄,以指定其為何種情境之對話之發言「請問 您需要什麼服務?」。一旦人對此回答以「我要找暑假的旅 遊去處。」等,即依輸入之「旅遊去處」、「找」等目錄 而指定「旅行介紹」之情境。另,在使用者之回答中,舉 例言之,包含「當地人員的應對態度很屋」、「導遊 3E稱職。」「飯店航辦與契約不符。」等否定性之單詞(在 文章中已以底線標出)在内時,則指定「投訴」之情境。如 上所述’本實施形態具有丄特徵,即,當人就多個情境(話 題)其中之一發言後,藉搜尋其中所包含作為目錄之單詞, 即可指定情境(話題)。即,藉搜尋作為事先已設定之目錄 之單詞,即可由眾多情境中選出特定之一種情境。 在快取記憶體之記錄程序(步驟1 3)中,CPU則依據目 錄而由記憶體叫出受指定之旅行介紹之情境(話題)之關連 式資料庫而將之記錄於快取記憶體中,且使其呈可自由产改 寫之狀態。 所謂旅行之情境綱目係由諸如「目的地」、「目的」、 房 「曰數」、「出發日(時)」、「人數」、「同行者之明細1、 「預异」、「指定之航空公司I 、「指定之飯店 、「 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公慶) -32- 五、 發明説明(3〇 間的種類」、「有無 護昭 Γ β兩… 」、有無自費活動」、「有無 此、 兩要簽證」、「付費方式」等所構成。因 貝訊項之有無判斷程序(步驟14)中。判 可=疋記錄所需之資訊項是否已完全齊備。㈣,以 1連串之β問應¥找出所欲記錄之諮詢模式,開始進 仃用以徵詢欠缺之資訊項之對話。 Β ‘本貫例中,電腦提出「請問您的目的地是?」之問 σ而使用者對此回| Γ英國或美國吧。」。電腦則由「英 國或美國吧。」之回答推知使用者尚未確定目的地,而轉 換為可加速確定目的地之建議模式。本發明即係根據人與 人之對話中,交替使用該諮詢模式與建議模式而發展對話 之常識,而將之初次應用於電腦與人間之對話中者。在本 實施例中,詢問旅行之目的後,即以使用者之回答「家族 敵仃」作為資訊項而推薦一旅遊資訊。由「C :您可考慮 佛羅里達的奥蘭多。」至「G:那就參加這個行程。」構 成了一次常式。接著,選定套裝行程後,即返回原本之旅 行介紹情境而繼續對話。 然後’依使用者所選定之套裝行程,「目的地」、 ---------¢------IX------線 (讀先閱請背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 曰 數 費用」、「航空公司」、「飯店、r 有無附餐 亦可確定,而僅須以後續之對話就其餘之方案加以徵詢 在本實施例中,尚未確認的有「出發日」及「人數」。 此,在徵詢程序(步驟1 5)中,即進行詢問「請問您希望 曰出發?」,而得到「7月1 8曰。」之回答。 本實施例其次將於轉換程序(步驟17)中,以「家族 因何 -33- 經濟部智慧財產局員工消費合作社印製 4736?{〇 ^^-—---- B7__ 五、發明説明(W ) $關鍵㈣轉換為有關「不占床計費」之小情境。具體而 言,詢問家族之組成、年齡後,即進行有關「不占床計費· 之呪明,以詢問是否接受該費率。假若,第4圖所示之對話 中第6仃之回答為諸如「新婚旅行」時,則可以「新婚」為 關鍵巧而轉換為該小情境。舉例言之,可提出由機場至飯 店之巴士接送、蜜月套房之特別晚餐、臨水鳥瞰套房等用 以增添_新婚氣氛之各種話題。 又’當已說出之輸入聲音中包含有用以指定記錄所需 之賁訊項時,則使用該記錄之資訊項,依對話順序而使程 式進行。通常,將依次向使用者確認已輸入之綱目而進行 之。另,在本實施例中,使用者正進行表達「要去佛羅里 達州奥蘭多之10天9夜套裝行程」之意。然而,其是否表達 了已確涊該套裝行程之所有條件之意,則可能於將來發生 紛爭時衍生問題。因此,就「目的地」、「日數」、「費 用」等已默許之事項亦宜加以明白確認。舉例言之,確認 「目的地」係奥蘭多(包含往NANSA之當曰來回巴士旅行) 與邁阿密,分別住6夜與2夜,交通需費3曰。 本樣態之特徵在可於上述對話中,提取出現於該對話 之主題’而藉對該主題深入探索或言及其他多樣性等而發 展對話。即,當預定之方案及/或元組成為話題時,可叫*出 用以規定小情境之關連式資料庫以作為次常式,而依項目 順序使程式進行。上述例子之特徵則在諸如當使用者言及 旅行之「目的」時,可控制話題以將之轉換為有關「家族 旅行」之小情境。 本纸張尺度適用巾國國家標準(CNS ) M規格(21〇><297公廣) ' --- -34- —-------批衣------、玎------0 (請先閲讀背面之注意事項再填寫本頁)-31 · 47367b A7 Printed by B7 Invention Cooperative of Employees ’Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs (29) In the memory of the brain, there are recorded dialogue sequences and project sequences that have been set to take the two topics and programs as topics, and Each program and item becomes the formula of the wording after the topic. Figure 4 shows an example of the dialogue when a computer introduces a person to a travel. In the directory inquiry program (step 12), first, the computer is used to tell the person Ask the place_wish list to specify the context in which it speaks. "What service do you need?" Once someone answers this with "I want to find a place to travel in the summer vacation." "," "Find" and other categories to specify the "travel introduction" situation. In addition, in the user's response, for example, "the local staff responded very well", "guide 3E competent." "Hotel airline If it does not agree with the contract. "And other negative words (marked with an underline in the article), the situation of" complaint "is designated. As described above, 'this embodiment has the characteristics That is, when a person speaks on one of multiple situations (topics), the situation (topic) can be specified by searching for words contained in it as a directory. That is, by searching for words as a directory that has been set in advance, Among the many scenarios, a specific one is selected. In the recording procedure of the cache memory (steps 13), the CPU calls a relational database of the situations (topics) specified by the travel according to the directory, and Record it in cache memory and make it freely rewritable. The so-called travel scenario outline is composed of "destination", "purpose", "room", "departure date (hour)" "" "Number of people", "Details of travellers 1," Pre-difference "," Designated airlines I, "Designated hotels," This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 public celebration) -32 -V. Description of the invention (30 types), "whether or not there are two or more ...", whether there are self-funded activities "," whether or not, two visas "," payment methods ", etc. Is there a judgment procedure (step 14). Judgment = 是否 Whether the information items required for recording are fully prepared. ㈣, use a series of β questions ¥ to find the consultation mode of the desired record, and start to use it for consultation Dialogue of missing information items. Β In this example, the computer asks "What is your destination?" Σ and the user responds to it | Γ United Kingdom or the United States. "The computer consists of" the United Kingdom or the United States. " The answer infers that the user has not yet determined the destination, and has switched to a suggested mode that can accelerate the determination of the destination. The present invention is based on the common knowledge of dialogue between people and the use of the consulting mode and the suggestion mode to develop a dialogue In the present embodiment, after inquiring the purpose of travel, the user's answer "family rival" is used as an information item to recommend travel information. From "C: You can consider Orlando, Florida." To "G: Then join this trip." It is a routine. Then, after choosing the package itinerary, you will return to the original journey to introduce the situation and continue the conversation. Then 'according to the package itinerary selected by the user, the "destination", --------- ¢ ------ IX ------ line (read first, please read the notes on the back, then (Fill in this page) Printed by the Consumers ’Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs,“ Airline ”,“ Hotels, and the presence or absence of meals can also be determined, and only the follow-up dialogue is needed to consult the remaining plans here. In the embodiment, "departure date" and "number of people" have not been confirmed. Therefore, in the inquiry procedure (step 15), the question "Do you want to leave?" Was asked, and the answer was "July 18". This example will be followed by the conversion process (step 17) with "Family Why -33-Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 4736? {〇 ^^ --------- B7__ V. Description of the invention ( W) $ Key ㈣ is converted into a small context about "no bed charge". Specifically, after asking about the composition and age of the family, a question about “no charge for the bed will be made” to ask whether to accept the rate. If the answer in the dialogue shown in Figure 4 is 6 Such as "new wedding trip", the "new wedding" can be transformed into the small situation. For example, various topics such as bus transfers from the airport to the restaurant, special dinners for honeymoon suites, and waterfront bird's-eye view suites can be added to enhance the wedding atmosphere. Also, when the spoken input sound contains information items necessary to specify the recording, the information items of the recording are used to make the procedure proceed according to the dialogue order. Normally, the user will sequentially confirm the entered schema. In addition, in this embodiment, the user is expressing the meaning of "a 10-day, 9-night package trip to Orlando, Florida". However, whether it expresses that all the conditions of the package itinerary has been confirmed may cause problems in the future when disputes occur. Therefore, tacit approval of "destination", "days", "costs", etc. should also be clearly confirmed. For example, confirm that the "destination" is Orlando (including the round-trip bus trip to NANSA) and Miami, and stay 6 nights and 2 nights respectively, and the transportation fee is 3 days. The features of this aspect can be used in the above-mentioned dialogue to extract the topics that appear in the dialogue 'and develop the dialogue by exploring the theme in depth or speaking about other diversity. That is, when a predetermined scheme and / or meta-composition is a topic, it can be called * out a relational database for specifying a small context as a subroutine, and the program is performed according to the order of items. The features of the above example are, for example, when the user talks about the "purpose" of the trip, the topic can be controlled to transform it into a small context about "family travel". This paper size is applicable to the national standard (CNS) M specification of towels (21〇 > < 297). '--- -34- --------- batch of clothes ------, 玎------ 0 (Please read the notes on the back before filling this page)
發明説明(32 經濟部智慧財產局員工消費合作社印製 在 豕族旅行」之小_ p 士 成之構造事例及其内容之將使用由諸多項目所構 以進行與人之對話。對話中内;=構成之關連式資料庫 '知旅订之項目有「家族之明細」、「孩童之性 」、孩童之年齡」、「是否採用不占床計費方式」、 「人數」等。上述之例子則在「請說明您、家人的資料 請說-明孩童的性別及年齡。」後,說明不占床計費方式, ^續問「請問您的決定是?」而轉料家族旅行之小話題。 旦就小情境之關連式資料庫中之所有構造事例或預定之 項目完成項目順序,即在餘式進行程序(步驟18)中回復 為對話順序而使其餘程式進行。 圖示之較佳實施例具有一特徵,即,用以規定小情境 之構造事例係指對話例。而由以上述程序集成之諸多内容 資訊所構成之小情境亦可記錄於記憶體中以作為對話例。 因此,下次當同一小情境成為話題時,即可根據該對話例 控制對話而使其依項目順序進行。藉逐漸累積對話例,即 可使電腦在詢問之方式、推銷預定之旅程等各方面熟能生 巧。 小情境可構築成無限之階層。即,一小情境可具有居 其下位之小情境,又,該居於下位之小情境又可具有更 其下位之小情境。藉此,電腦與人間之對話之多樣性可 限延伸,而完全消除所謂單純運用預備會話之與電腦對話 的獨特單調性。 其次,參照第5圖,就用以實施第3圖所示之主題對話 別 厂 居 Μ ^II線 (請先閱讀背面之注意事項再填寫本頁) 本紙張尺度適用中國國家標準(CNS ) Α4規格(210Χ297公釐) •35- 4736?{〇 A7 經濟部智慧財產局員工消費合作社印製 B7五、發明説明(33 ) " "" 方法之系統加以詳細說明。 由於本實施例之系統除主電腦9之構造外皆與第2圖所 示之主題對話系統相同,故僅就相異之構造加以說明。又, 在說明中,與第2圖相同之構成要素則使用與第2圖中相同 之參照號碼。 在本實施例中,主電腦9之記憶體11内設有一關連式資 料庫部—1 U,一用以記錄已設定以何順序將各方案加入話題 之程式之對話順序部nb,及,一用以記錄已設定各方案成 為話題時之措辭等之程式之措辭記錄部丨lc。關連式資料庠 部11 a並分割成區段11 aa〜1丨an而可儲存諸多不同情境(話 題)之關連式資料庫。在圖示之較佳實施例中,各關連式資 料庫中皆登錄有預先選出之可作為與他者識別之目錄之一 個或多個單詞。而,藉搜尋該單詞,即可指定一關連式資 料庫。另,與第2圖所示之1例子相同,各關連式資料庫亦皆 為綱目與元組所構成者。 記憶體11中並設有用以規定與預定之方案及/或元組 相關連之小情境之一個或多個關連式資料庫部〗I d。而當其 與資料庫部11 a同樣具有多個用以規定小情境之關連式資 料庫時,關連式資料庫部lid亦可分割成區部11(^〜11(1111。 且,用以規定小情境之關連式資料庫亦為由諸多項目所1構 成之構造事例(相當於綱目)及其等之内容之内容事例(相 當於元組)所構成者。又,用以規定各小情境之關連式資料 庫並須事先設定當用以規定情境之關連式資料庫中之何元 組出現時應轉換為該小情境。 本紙張尺度適用中關家標準(CNS ) A4規格(21GX297公釐) " --- -36- 裝 訂 線 (請先閱讀背面之注意事項再填寫本頁} 經濟部智慧財產局員工消費合作社印製 473671) A7 _____Β7 ___ 五、發明説明(34 ) 又,記憶體11中更設有一用以記錄以各項目為話題之 次序之項目順序之項目順序部11 e,及一用以記錄已設定各 項目成為話題後之措辭等之程式之措辭記錄部llf。 卢,主電腦9在記憶體11之外,並設有可叫出依據目錄 而受指定之關連式資料庫而記錄之,且使其呈可自由改寫 之狀態之快取記憶體15。 C?U 13與第2圖中之系統同樣設有一資訊項有無判斷 控制裝置13a,一必要資訊項徵詢控制裝置i3b,及,一程 式進行控制裝置13c。且,CPU13内並設有一由電腦向人詢 問所欲之目錄,以指定其為何種情境之對話之目錄詢問控 制裝置13d,及,一可依照輸入之目錄指定情境,再由記憶 體叫出被指定之關連式資料庫而將之記錄於快取記憶體 中’且使其呈可自由改寫之狀態之快取記憶體記錄控制裝 置13e。進而,CPU13内更設有一用以於預定方案及/或元 組成為話題時,叫出用以規定小情境之關連式資料庫以作 為次常式’而依照項目順序使程式進行之次常式進行控制 裝置13f,及,一用以於項目順序結束時,回復為對話順序 而使其餘程式進行之回復對話順序控制裝置13g。 在圖示之較佳實施例中’必要資訊項徵詢控制裝置1 3 b 係可將方案名稱加入徵詢句以進行詢問,而藉此進行徵詢 欠缺之資訊項者。 第6圖係本發明之第3樣態中之主題對話方法之一實施 例之流程圖。 第7圖係顯示用以實施第6圖所示之主題對話方法之系 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐〉 裝------訂------線 (請先閱讀背面之注意事項再填寫本頁) -37- 4736?{〇 五、發明説明(π 統中主電腦9之一實施例之方塊圖。 本發明之第3樣態中之主題對話方法與第 對話方法相較,其不同之處在於其使用了記錄料有由屬 性與專級所構成之知識化資料之知識化資料庫,以替代關 連式資料庫。此處之知識化資料則係指由諸多屬性/等級組 所構成之資料。而’屬性與關連式資料庫之方案相較由 於其免名稱(相當於方案)之外同時亦記載了限制條件故 可大幅縮短檢索時間且大幅提昇檢索之精確度。 本樣態中之主題對話方法大致包含有··一知識化單詞 辭典作絲序(步職),—知識化轉庫.程式記錄程序 (步驟S22),一資訊項有無判斷程序(步驟23),一當輸入之 訂 聲音中是否包含用以指定知識化資料所需之資訊項之判斷 J21結果為「否」時之必要資訊項徵詢程序(步驟2句,及, 一當判斷J21結果為「是」,時,或藉徵詢程序(步驟2句而齊 備指定知識化資料所需之資訊項後之程式進行程序(步驟 S25) 〇 線 在知識化單詞辭典作成程序(步驟21)中,可將出現於 屬預定對話所進行之情境(話題)之資料中之單詞,依各名 稱加以分類,並以作為該分類名之名稱與該名稱之類型、 大小及可取得之範圍等條件為其屬性,且以該名稱之值彳作 為等級之知識化單詞加以累積。 在知識化資料庫·程式記錄程序(步驟22)中,則可於 電月包之δ己憶體中記錄包含相同屬性組之多個或單個知識化 貧料庫所具有之諸多屬性/等級組,同時記錄已設定以各屬 本紙張尺度適用中國國家標準(CNS ) Α4規格(2Κ)χ297公瘦 4736?(〇 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明説明(36 性為話題時之次序的對話順序及各屬性成為話題後之措辭 等之程式。且,如後述部分之詳細說明,知識化資料並可 藉自既存之關連式資料庫及物件導向資料庫進行簡單之置 換而輊易加以構築。又,藉實施本樣態中之主題對話方法 而得到之屬性及等級則可直接儲存於本知識化資料庫中以 作為對話例。藉如此累積對話例,即可使電腦習得可謂對 話訣竅之說話技巧。 在資訊項有無判斷程序(步驟23)中,電腦則可利用上 述之知識化單詞辭典,對使用者所輸入之聲音進行單詞分 解及/或文章解析。而,藉使用知識化資料,即可達到正確 且省時地完成聲音辨識之效果。 在必要資訊項徵詢程序(步驟24)中,電腦則可判斷有 無與各屬性相對應之資訊項,尤其是與其名稱相對應者。 在必要資訊項徵詢程序(步驟24)中,亦可於輸入之聲 音中未包含用以指定知識化資料所需之資訊項時,就欠缺 之資訊項徵詢使用者,以取得該所需之全部資訊項目。 在程式進行程序(步驟25)中,則可於輸入之聲音中包 含用以指定知識化資料所需之資訊項時(包含已藉步驟24 而取得必要資訊項時),使用該知識化資料之資訊項,依照 對話順序使程式進行。 本樣態係使用了具通用性且可正確而迅速地辨識輸入 之聲音之知識化資料庫,以替代上述第一樣態中所使用之 關連式資料庫者。因此,在本樣態中,電腦可使用上述之 儲存於知識化資料庫中之知識化資料以與人對話。 表紙張尺度適用中國國家標準(CNS ) M規格(21Gx297公羡 I I - I I I I 批水 111 : I In ^ 111111 ^ (請先閱讀背面之注意事項再填寫本頁) -39- B7 五、發明説明(37) 其次,參照第7圖,就用以實施第6圖所示之主題對話 方法之系統加以詳細說明。 f請先閱讀背面之注意事項再填寫本頁} 由於本實施例中之系統除主電腦9之構造外皆與第2圖 所不之主題對話系統相同,故僅就不同之構造加以說明。 又,在說明中,與第2圖相同之構成要素則使用與第2圖中 相同之參照號碼。 免本實施例中,主電腦9之記憶體11與第2圖中之系統 同樣設有一用以記錄已設定以何順序將各屬性加入話題之 私式之對話順序部丨lb,及,一用以記錄已設定各屬性成為 話題後之措辭等之程式之措辭記錄部丨丨c。且,記憶體j j 中並設有一用以記錄儲存知識化單詞之知識化單詞辭典部 1 lg,及,一用以記錄儲存後述部分之詳細說明中之知識化 資料之知識化資料庫部丨丨h。 經濟部智慧財產局員工消費合作社印製 又,主電腦9之CPU13並與第2圖中之系統同樣設有一 資訊項判斷控制裝置13a,係可利用知識化單詞辭典對使用 者輸入之聲音進行單詞分解及/或文章解析,以判斷有無 與各屬性相對應之資訊項者;一必要資訊項徵詢控制裝置 13b’係用以於輸入之聲音中未包含用以指定知識化資料所 需之資訊項時,就欠缺之資訊項徵詢使用者,以取得該所 需之全部資訊項目者;及,一程式進行控制裝置丨3c,係<用 以於輸入之聲音中包含有用以指定知識化資料所需之資訊 項時,使用該知識化資料之資訊項,依照對話順序使程式 進行者。且,CPU13内更設有一知識化單詞辭典作成控制 裝置13h ’係用以作成知識化單詞辭典者。 本紙張尺度適用中國國家標準(CNS ) A4規格(2][〇><297公羡) 經濟部智慧財產局員工消費合作社印製 47367^0 五、發明制(38 ) " ^--- ^ 第8圖係本發明之第4樣態中之主題對話方法之一實施 例之流程圖。 ' 第9圖係顯示用以實施第8圖所示之主題對話方法々争 統中丰電腦9之一實施例之方塊圖。 … ▲本發明之第4樣態中之主題對話方法與第2樣態之主題 對居方法相較,其不同之處在於其使用了記錄餘存有由屬 性與等一級所構成之知識化資料之知識化資料庫,以替代關 連式資料庫。即,其未使用對應同一綱目儲存有諸多記錄 之關連式資料庫,而使用了各由獨特之屬性與等級所構成 之知識化資料。因此,為識別各知識化資料與其他大量之 知識化資料,必須對其授與目錄,或適切地選擇已儲存以 作為等級之單詞之複數而以之作為目錄。其次,透過與詢 問者之對話,即可指定情境(話題ρ因此,在欲指定一知 識化資料時,將先判斷輸入之聲音中是否包含必要資訊 項,若不包含,則藉進一步之對話徵詢不足之資訊項,而 此點亦與第2樣態之主題對話方法不同。 本樣態中之主題對話方法大致包含有:一知識化單詞 辭典作成程序(步驟3 1); —知識化資料庫·程式記錄程序 (步驟32),· 一目錄詢問程序(步驟33); 一資訊項有無判斷程 序(步驟3 4),一當輸入之聲音中是否包含用以指定知識化 資料所需之資訊項之判斷J31結果為「否」時之必要資訊項 徵詢程序(步驟35” 一當判斷J31結果為「是」時,或藉徵 詢程序(步驟3 5)齊備用以指定知識化資料所需之資訊項後 之相關知識化資料之快取記憶體記錄程序(步驟36); 一佐 本紙張尺度適用中國國家標準(CNS ) Α4規格(210Χ297公釐) ' ^ -- -41· ---------裝------訂 線 (請先閱讀背面之注意事項再填寫本頁) 47 367{〇 經濟部智慧財產局員工消費合作社印製 A7 B7 五、發明説明(39 ) 照對話順序進行之程式進行程序(步驟37); 一次常式程式 進行程序(步驟38)及一用以回復對話順序之程序(步驟39)。 在知識化單詞辭典作成程序(步驟3丨)中,可將屬多個 或其中之特定情境(話題)之各資料中出現之單詞,依各名 稱加以分類’並以作為該分類名之名稱與該名稱之類型、 大小及可取得之範圍等條件為其屬性,且以該名稱之值作 為等級_之知識化單詞加以記錄儲存,而作成知識化單詞辭 典。 在知識化資料庫·程式記錄程序(步驟32)中,可於電 腦之記憶體中記錄具有可識別各情境(話題)之目錄之諸多 知識化資料之屬性及等級,以及,由用以規定與全部或預 疋之屬性及/或等級相關連之小情境之知識化資料之諸多 項目所構成之構造事例及其内容之内容事例。且,其中亦 可記錄已設定以各屬性及項目為話題時之次序的對話順序 及項目順序,以及,各屬性及項目成為話題後之措辭等之 程式。 在目錄詢問程序(步驟33)中,則可由電腦向人詢問所 欲之目錄’以指定其為何種情境之對話。 在資訊項有無判斷程序(步驟34)中,則可利用知識化 單詞辭典對使用者所輸入之聲音進行單詞分解及/或文《章 解析,以判斷有無用以指;^知識化資料所需之全部資 而,當該輸入之聲音中未包含、 不匕3用以指定知識化資料所 需之資訊項時’即可在必要資訊項徵詢程序(步驟叫中, 就欠缺之資訊項,即,本樣態中作為目錄之單詞,徵詢使 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公慶 -------—------π------it (請先閲讀背面之注意事項再填寫本頁) -42- B7 五、發明説明(4〇 ) 用者以取得該所需之全部資訊項目。 (請先閲讀背面之注意事項再填寫本頁) 另’當該輸入之聲音中包含用以指定記錄所需之資訊 項時’則可在快取記憶體記錄程序(步驟36)中,依照目錄 指定情境,再由該記憶體叫出所指定之知識化資料庫而將 之§己錄於快取記憶體中,並使其呈可自由改寫之狀態。 在程式進行程序(步驟37)中,電腦則將使用該知識化 資料之·資訊項,進行依循對話順序之程式。 在次常式程式進行程序(步驟3 8)中,當預定之屬性及/ 或等級在依循對話順序之程式進行中成為話題時,則可叫 出以該屬性及/或等級為用以與其:^小情境識別之目錄之 知識化資料以作為次常式,而依照項目順序使程式進行。 項目順序結束後,即在用以回復對話順序之程序(步驟 39)中,回復為對話順序而使其餘程式進行。 經濟部智慧財產局員工消費合作社印製 由於本樣態中所使用之知識化資料庫係可藉自既存之 關連式資料庫及物件導向資料庫進行簡單置換而輕易加以 構築者’故適切地選擇可識別各知識化資料與其他知識化 >料之目錄’即可利用現存數量魔大之資訊。即,電腦將 可就可能為人引為話題之所有情境進行對話,而進一步達 到電腦之人性化。 其-人,參照第9圖’就用以實施第8圖所示之主題對話 方法之系統加以詳細說明。 由於本實施例中之系統除主電腦9之構造外皆與第2圖 所示之主題對話系統相同,故僅就不同之構造加以說明。 又,在說明中,與第2圖相同之構成要素則使用與第2圖中 本紙張尺度適用中國國家標準(CNS ) Α4規格(210X297公釐) 經濟部智慧財產局員工消費合作社印製 473671) A7 — — B7 五、發明説明(41 ) 相同之參照號碼。 /在本實施例中,主電腦9之記憶體丨〗内設有一用以儲存 後述部分之詳細說明中之大量知識化資料之知識化資料庫 #8a,及用以規定與全部或預定之情境之屬,隹及/或等 級相關連之小情境之知識化資料庫部8d。1,已對知識化 資料庫授與可識別各情境之目錄或已選定作為目錄之單 司又_與第2圖中之系統相同,記憶趙1 j内並設有用以記 錄已設定以何順序將各屬性加入話題之程式之對話順序部 肋,及,用以記錄已設定各屬性成為話題後之措辭等之程 式之措辭記錄部8c。 進而,記憶體11内更設有一用以記錄以各項目(相當於 用以規定小情境之知識化資料之屬性)為話題之次序之項 目順序(相當於用以規定小情境之知識化資料之對話順序) 之項目順序部n 一用以記錄已設定各項目成為話題 後之措辭等之程式之措辭記錄部8fe此外,記憶體丨丨内又 設有用α記錄儲存與第7圖所示之主題對話系統相同之知 識化單詞之知識化單詞辭典部8g。 而,主電腦9在記憶體〗!以外,並設有可於叫出依據目 錄而受指定之知識化資料後,加以記錄、改寫之快取記憶 體17。而,已設定各知識化資料、對話順序及各屬性成♦為 話題後之措辭等之程式則宜彼此相關而儲存。同樣地已 設定用以規定小情境之知識化資料、項目順序及各構造事 例成為話題後之措辭等之程式亦宜彼此相關*儲存。因 此,當欲將受指定之知識化資料或用以規定小情境之知識 本紙張尺度適用中國國家標準(CNS ) A4規格(210X29*7公釐) ------ •44·Description of the Invention (32 Employees' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, printed on the Dai nationality travel ", the small _ p Shi Cheng ’s construction examples and their content will use a number of items to construct dialogue with people. In the dialogue; = Constituted relational database formed by the Zhilv Travel Project includes "family details", "children's sex", "children's age", "whether non-occupancy charge method is used", "number of people", etc. The above examples Then after "Please explain the information about you and your family-please indicate the gender and age of the child.", Explain the cost of not occupying a bed, ^ Continue to "Is your decision?" And turn to the small topic of family travel Once the project sequence is completed for all the construction cases or predetermined items in the small-scenario relational database, that is, the rest of the procedure is reverted to the dialog sequence in the residual procedure (step 18) and the remaining procedures are performed. The example has a feature that the construction example used to specify the small context refers to the dialogue example, and the small context composed of a lot of content information integrated by the above procedure can also be recorded in the memory as a dialog Therefore, next time when the same small situation becomes a topic, you can control the dialogue according to the dialogue example so that it will proceed in the order of items. By gradually accumulating dialogue examples, you can make the computer in the way of inquiries and sales of scheduled journeys. Aspects make perfect. Small situations can be constructed into infinite strata. That is, a small situation can have a small situation below it, and a small situation that can reside below it can also have a small situation below it. By this, the computer and the world The diversity of the dialogue can be limited, and the unique monotony of the dialogue with the computer, which simply uses preparatory conversation, can be completely eliminated. Second, referring to Figure 5, it is used to implement the theme dialogue shown in Figure 3. Line II (please read the precautions on the back before filling this page) This paper size applies the Chinese National Standard (CNS) Α4 size (210 × 297 mm) • 35- 4736? {〇A7 Printed by the Employees ’Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs B7 V. Invention description (33) The system of the method is explained in detail. Because the system of this embodiment is the same as that shown in FIG. 2 except for the structure of the host computer 9. The subject dialogue system shown is the same, so only the different structures will be described. In the description, the same reference numerals as those in FIG. 2 are used for the same constituent elements as in FIG. 2. In this embodiment, the main The memory 11 of the computer 9 is provided with a related database part—1 U, a dialogue order part nb for recording a program in which order has been added to the topic, and a record for each set program. The wording record section of the formula such as the wording at the time of the topic 丨 lc. The related data section 11a is divided into sections 11aa ~ 1 丨 an to store a related database of many different situations (topics). In the figure In the illustrated preferred embodiment, one or more words pre-selected as a directory that can be recognized by others are registered in each related database. Instead, by searching for that word, you can specify a connected database. In addition, like the example shown in Fig. 2, each related database is composed of outlines and tuples. The memory 11 is also provided with one or more related database parts Id for specifying a small context associated with a predetermined scheme and / or tuple. And when it has a plurality of related databases that are used to specify a small situation, like the database department 11 a, the related database department lid can also be divided into districts 11 (^ ~ 11 (1111.) And used to specify The relational database of small contexts is also composed of structural examples (equivalent to outlines) composed of many items1 and content examples (equivalent to tuples) of the content. Also, it is used to specify the The relational database must be set in advance when any tuple in the relational database used to specify the situation should be converted to the small situation. This paper size applies the Zhongguanjia Standard (CNS) A4 specification (21GX297 mm) " --- -36- gutter (please read the notes on the back before filling out this page) printed by the Employees' Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs 473671 A7 _____ Β7 ___ V. Description of the invention (34) Also, memory 11 In addition, there is an item ordering section 11e for recording the order of items with the order of each item as the topic, and a wording recording section 11f for recording the programs that have set the wording after each item becomes the topic. Lu, Zhu Computer 9 in memory Besides the body 11, there is a cache memory 15 which can call out a related database which is designated according to the catalogue and records it in a freely rewriteable state. C? U 13 and Fig. 2 The system is also provided with an information item judgment control device 13a, a necessary information item inquiry control device i3b, and a program execution control device 13c. Furthermore, a CPU 13 is also provided with a computer to ask people for a desired directory to specify It is a directory inquiry control device 13d of what kind of context it is, and, a context can be specified in accordance with the entered directory, and then the designated relational database is called from the memory and recorded in the cache memory 'and used It is a cache memory recording control device 13e that can be rewritten freely. Furthermore, the CPU 13 is provided with a relational database for defining small scenarios when a predetermined scheme and / or meta-composition is the topic. As a subroutine, the subroutine execution control device 13f for program execution in accordance with the order of items, and a reply pair for restoring the rest of the programs to the dialogue order at the end of the item sequence. Sequence control device 13g. In the preferred embodiment shown in the figure, the 'necessary information item inquiry control device 1 3 b is a person who can add the scheme name to the inquiry sentence to make an inquiry, and thereby conduct the inquiry for the missing information item. FIG. 6 FIG. 7 is a flowchart of an embodiment of the theme dialogue method in the third aspect of the present invention. FIG. 7 shows the system used to implement the theme dialogue method shown in FIG. 6. The paper standard is applicable to the Chinese National Standard (CNS). A4 specifications (210X297 mm) ------- order --- line (please read the precautions on the back before filling out this page) -37- 4736? {〇 五 、 Explanation of invention (π 中 中Block diagram of an embodiment of the host computer 9. The subject dialogue method in the third aspect of the present invention is compared with the second dialogue method in that it uses a knowledgeable database that records knowledgeable data composed of attributes and specializations to replace the relationship. Database. Knowledge data here refers to data composed of many attribute / level groups. Compared with the scheme of the 'attribute' and related database, because of its exemption (equivalent to the scheme), it also records the restriction conditions, which can greatly reduce the retrieval time and greatly improve the accuracy of retrieval. The theme dialogue method in this aspect roughly includes a knowledge word dictionary as a silk order (step), a knowledge transfer library, a program recording procedure (step S22), and an information item judgment process (step 23) , The necessary information item inquiry procedure when the input order voice contains the information item required to specify the knowledge item to determine the J21 result is "No" (step 2 sentence, and, when the result of the judgment J21 is " "Yes", or through the inquiry procedure (step 2) and the procedure after specifying the information items required for the intellectualized data (step S25). In the knowledge word dictionary creation procedure (step 21), Words that appear in the data that belongs to the context (topic) in which the dialogue is scheduled are classified according to each name, and the attributes such as the name of the classification name and the type, size, and available range of the name are attributes, And the value of the name 彳 is used to accumulate knowledgeable words in the rank. In the knowledge database and program recording program (step 22), the same can be recorded in the δ self-memory body of the electric monthly bag. Multiple attribute / level groups of multiple or single knowledge-based lean materials in the attribute group. At the same time, the records have been set to apply the Chinese National Standard (CNS) Α4 specification (2Κ) χ297 male thin 4736? (〇 A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 5. The invention description (36 is the order of the dialogue when the topic is the topic, and the formula of the wording after each attribute becomes the topic.) Data can be easily constructed by simply replacing existing relational database and object-oriented database. Also, the attributes and levels obtained by implementing the theme dialogue method in this aspect can be directly stored in this The knowledge database is used as an example of dialogue. By accumulating the dialogue examples in this way, the computer can acquire the speaking skills that can be described as the know-how of the dialogue. In the information item judgment process (step 23), the computer can use the above-mentioned knowledge word dictionary , To perform word decomposition and / or article analysis on the voice input by the user. By using knowledgeable data, you can achieve correct and time-saving Complete the effect of sound recognition. In the necessary information item inquiry procedure (step 24), the computer can determine whether there is an information item corresponding to each attribute, especially the name corresponding to it. The necessary information item inquiry procedure (step 24) In addition, when the inputted voice does not include the information items required to specify the intellectualized data, the user is consulted for the missing information items to obtain all the required information items. Perform the procedure in the program (step 25 ), When the input voice contains the information items required to specify the knowledgeable data (including when the necessary information items have been obtained through step 24), use the information items of the knowledgeable data in accordance with the dialog order This program uses a knowledgeable database that is universal and can correctly and quickly recognize the input voice, instead of the related database used in the first mode. Therefore, in this aspect, the computer can use the above-mentioned intellectualized data stored in the intellectualized database to talk to people. The paper size of the table is applicable to the Chinese National Standard (CNS) M specification (21Gx297 Public Envy II-IIII batch of water 111: I In ^ 111111 ^ (Please read the precautions on the back before filling out this page) -39- B7 V. Description of the invention ( 37) Secondly, referring to Fig. 7, the system for implementing the theme dialogue method shown in Fig. 6 will be described in detail. F Please read the precautions on the back before filling in this page} Since the system in this embodiment The structure of the computer 9 is the same as that of the theme dialogue system shown in FIG. 2, so only the different structure will be described. In the description, the same components as those in FIG. 2 are used as those in FIG. Reference number. In this embodiment, the memory 11 of the host computer 9 and the system in FIG. 2 are also provided with a private dialogue sequence unit for recording in which order the attributes are added to the topic. A word recording unit for recording the formulas that have been set for each attribute to become the topic, etc. c. Moreover, the memory jj is also provided with a knowledgeable word dictionary unit 1 lg for storing and storing knowledgeable words, And, one for The knowledge database is stored and stored in the detailed description of the later part. It is printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs and printed on the CPU13 of the host computer 9. The information item judgment control device 13a is a person who can use the knowledgeable word dictionary to perform word decomposition and / or article analysis on the voice input by the user to determine whether there is an information item corresponding to each attribute; a necessary information item consultation control device 13b 'It is used when the inputted voice does not contain the information items required to specify the intellectualized data, and the user is consulted for the lack of information items to obtain all the required information items; and, a program controls The device 丨 3c is used to make the program performer in accordance with the dialogue order when the inputted sound contains information items useful for specifying the knowledgeable data, and the information items are used in accordance with the dialogue order. A knowledgeable word dictionary creation control device 13h is provided to create a knowledgeable word dictionary. The paper standard is applicable to the Chinese National Standard (CNS) A 4 Specifications (2) [〇 > < 297 public envy) Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 47367 ^ 0 V. Invention System (38) " ^ --- ^ Figure 8 is the first of the invention A flowchart of one embodiment of the theme dialogue method in 4 aspects. 'FIG. 9 is a block diagram showing one embodiment of the Zhongfeng computer 9 for implementing the theme dialogue method shown in FIG. 8... ▲ The subject dialogue method in the fourth aspect of the present invention is compared with the subject opposition method in the second aspect. The difference is that it uses a record of the remaining knowledgeable data composed of attributes and equal levels. Knowledge-based databases to replace connected databases. That is, instead of using a connected database that stores many records corresponding to the same outline, it uses knowledgeable data each composed of unique attributes and levels. Therefore, in order to identify each intellectualized material and a large number of other intellectualized materials, a catalogue must be given to them, or a plurality of words which have been stored as a rank must be appropriately selected as a catalogue. Secondly, the situation can be specified through a dialogue with the interrogator (topic ρ. Therefore, when you want to specify a knowledgeable material, you will first determine whether the input voice contains the necessary information items. If it does not contain it, you can ask for further dialogue Insufficient information items, and this point is also different from the theme dialogue method of the second aspect. The theme dialogue method in this aspect roughly includes: a knowledge word dictionary creation process (step 31);-knowledge database Program recording procedure (step 32), a directory inquiry procedure (step 33); an information item judgment procedure (step 3 4), whether the input voice contains the information items required to specify the intellectualized data Necessary information item consultation procedure when the J31 result is "No" (Step 35) When the J31 result is "Yes", or the consultation process (Step 3 5) is available to specify the information required for knowledgeable data Cache memory recording procedure for related knowledge-based data after the item (step 36); The paper size applies the Chinese National Standard (CNS) A4 specification (210 × 297 mm) '^--41 · ----- ---- Equipment ------ Order the line (please read the precautions on the back before filling this page) 47 367 {〇 Printed by the Consumers ’Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 V. Description of Invention (39) Photo The procedure of the dialogue sequence (step 37); a routine procedure (step 38) and a procedure to restore the dialogue sequence (step 39). In the knowledge word dictionary creation procedure (step 3 丨), Words appearing in various materials that belong to multiple or specific situations (topics) can be classified according to each name 'and the conditions such as the name of the classification name and the type, size, and scope of the name can be obtained as Its attributes are recorded and stored with the value of the name as the knowledgeable word of rank_, and a knowledgeable word dictionary is created. It can be recorded in the computer's memory in the knowledge database and program recording program (step 32). Attributes and levels of knowledgeable data with a catalog that identifies each context (topic), and knowledgeable data used to specify small contexts associated with all or predicted attributes and / or levels Examples of the structure and content of many items. Also, it is possible to record the dialogue order and item order in which the order of each attribute and item has been set, and the wording after each attribute and item becomes the topic. In the directory inquiry procedure (step 33), the computer can ask people for the desired directory 'to specify the context of the dialogue. In the information item existence judgment procedure (step 34), knowledge can be used The word dictionary performs word decomposition and / or chapter analysis on the voice input by the user to determine whether it is used for reference; ^ All the resources required for knowledgeable data are not included in the input voice 3 When used to specify the information items required for knowledgeable data, you can consult the necessary information items (the step is called, the missing information items, that is, the words in this form as a catalog, and the consultation makes the paper size applicable) China National Standard (CNS) A4 Specification (210X297 Public Holiday --------------- π ------ it (Please read the precautions on the back before filling out this page) -42- B7 V. Description of Invention 4〇) with a person to get all the information required of the project. (Please read the precautions on the back before filling this page) In addition, when the input sound contains the information items required to specify the record, you can use the directory in the cache memory recording procedure (step 36) Specify the situation, and then call out the specified knowledgeable database from the memory, and record it in the cache memory, and make it rewritable. In the program execution process (step 37), the computer will use the information items of the knowledge data to carry out the program following the dialogue order. In the subroutine program execution process (steps 3 to 8), when the predetermined attribute and / or level becomes a topic in the process of following the dialog sequence, the attribute and / or level can be called to use it: ^ The knowledgeable data of the catalog of small context recognition is used as a subroutine, and the program is performed according to the order of items. After the item sequence is completed, in the procedure (step 39) for restoring the conversation sequence, the rest of the procedures are resumed for the conversation sequence. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs, because the knowledge-based database used in this mode can be easily constructed by simply replacing the existing connected database and the object-oriented database. A catalog that can identify each knowledgeable material and other knowledgeable materials> can use the existing large amount of information. That is, the computer will be able to talk about all situations that may be a topic for humans, and further the humanization of the computer. The person is described in detail with reference to Fig. 9 'for a system for implementing the theme dialogue method shown in Fig. 8. Since the system in this embodiment is the same as the subject dialogue system shown in FIG. 2 except for the structure of the host computer 9, only the different structures will be described. In the description, the same constituent elements as those in Figure 2 are printed using the Chinese paper standard (CNS) A4 specification (210X297 mm) and the paper size in Figure 2 (printed by the Consumer Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs 473671) A7 — — B7 V. Description of Invention (41) Same reference number. / In this embodiment, the memory of the host computer 9 is provided with a knowledgeable database # 8a for storing a large amount of knowledgeable information in the detailed description of the later-mentioned part, and a situation for specifying and all or predetermined It belongs to the knowledge database part 8d of the mini-situation related to the level and / or level. 1. The knowledge database has been given a catalog that can identify each situation or a single company that has been selected as a catalog. It is the same as the system in Figure 2. It is stored in Zhao 1j and is used to record the order in which it has been set. A dialogue sequence part of a formula for adding each attribute to a topic, and a wording recording part 8c for recording a formula in which the wording of each attribute has been set as a topic. Furthermore, a sequence of items (corresponding to the intellectualized data for specifying the small context) is recorded in the memory 11 to record the order of the topics (corresponding to the attributes of the intellectualized data for specifying the small context). Dialogue order) Item ordering section n A wording recording section 8fe for recording the programs that have set the wording of each item after becoming a topic, etc. In addition, the memory is also provided with α record storage and the subject shown in Figure 7 The knowledgeable word dictionary unit 8g of the knowledgeable word having the same dialogue system. However, the host computer 9 is in memory! In addition, there is a cache memory 17 which can be recorded and rewritten after calling out the intellectualized data specified according to the directory. However, the programs that have set each intellectualized data, the order of dialogue, and each attribute into the wording after the topic should be stored in relation to each other. Similarly, the procedures that have been set up to define the knowledgeable data of the small context, the order of the items, and the wording after each structural case becomes the topic should also be related to each other * and stored. Therefore, when you want to assign the designated knowledgeable information or the knowledge used to specify the small context, the paper size applies the Chinese National Standard (CNS) A4 specification (210X29 * 7 mm) ------ • 44 ·
----裝-- (請先閲讀背面之注意事項再填寫本頁) 、-|1 線 4736? 經濟部智慧財產局w工消費合作社印製 A7 B7 i、發明説明(42) 化資料記錄於該快取記憶體17中,且使其呈可自由改寫之 狀態時,亦將同時將已設定對話順序及各屬性成為話題後 之措辭等而與前二者相關之程式記錄於快取記憶體1 7内, 並使异呈可自由改寫之狀態。 另,主電腦9之CPU13内則設有一知識化單詞辭典作成 控制裝置14h,係用以作成與第7圖所示之主題對話系統相 同之知-識化單詞辭典者;一目錄詢問控制裝置14b,係由電 腦向人詢問所欲之目錄,以指定其為何種情境之對話者。 且,CPU13内並設有一資訊項判斷控制裝置14a,係可利用 知識化單詞辭典,對使用者回應目錄之詢問而輸入之聲音 進行單詞分解及/或文章解析,以判斷有無用以指定知識 化=貝料所需之全部資訊項者;及,一必要資訊項徵詢控制 裝置14d,係用以於輸入之聲音中未包含用以指定知識化資 料所需之資訊項時,就欠缺之資訊項徵詢使用者,以取得 該所需之全部資訊項目者。 進而,CPU13内更設有一快取記憶體記錄控制裝置 14以係當輸入之聲音中已包含用以指定知識化資料所需之 為訊項時,可依照目錄指定情境,再由記憶體叫出受指定 之知識化資料而將之記錄於快取記憶體中,且使其呈可自 由改寫之狀態者;及,一程式進行控制裝置〗4c,係可依照 預先設定之對話順序使程式進行者。 、 此外,CPU13内又設有一次常式進行控制裝置14f,係 當預定之屬性及/或等級於程式依照對話順序進行中成為 后題時,可叫出用以規定小情境之知識化資料(亦包含關連 本紙張尺度適用中國國家標準(CNS ) A4規格(21()><297公董 ---------裝------訂------線 (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 47367b A7 ------ B7__ 五、發明説明(43 ) 式 > 料庫)以作為次常式,而依照項目順序使程式進行者; 及,一回復對話順序控制裝置14g,係用以於項目順序結束 後’回復為對話順序而使其餘程式進行者。 其次,參照第10圖,就知識化資料庫之構築方法加以 詳細說明。 第10圖係用以說明本發明之第3及第4樣態中之主題對 話方法 / 斤使用之知識化資料庫之構築方法之流程圖。 第11圖係顯示知識化單詞之數例之表格。 參照號碼10係代表屬多種話題、業界、領域(與各話 題、業界、領域相關之資訊之集合稱為「情境」)之資料的 市σ者β右賣料為外語文書10a,則藉翻譯機12a翻譯成曰 文,再將之送至特殊化控制裝置20。若資料為一般文書, 則於OCRUb進行預處理,若資料為聲音,則藉聲音辨識軟 體12c等為人所熟知之方法進行預處理然後再將之送至特 殊化控制襄置20。另,右資料為各種電子資訊,則藉格 式轉換裝置12d將格式統一,再將之送至特殊化控制裝置 20 〇 立本樣態使用了知識化單詞辭典以進行單語分解及/或 文章解析,而將作為輸入之資訊之與人的對話及各種資料 庫特殊化。而,此處之知識化單詞辭典則係储存有依各《名 稱將單詞分類,並以作為該分類名之名稱與該名稱之類 型、大小及可取得之範圍等條件為其屬性,且以該名稱之 值作為等級之知識化單詞者。如同第4樣態當情境(話題) 涵括多種類時,即對屬特定情境(話題)之各資料收集單 i紙張尺度適用_家標準(CNS)公慶____.—__— - . 衣1T------.^ (請先閱讀背面之注意事項再填寫本頁) -46· ,的7丨0---- Equipment-(Please read the precautions on the back before filling in this page),-| 1 line 4736? Printed by A7 B7, Intellectual Property Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs, i. When the cache memory 17 is in a state where it can be freely rewritten, the programs related to the former two are recorded in the cache memory at the same time as the dialog sequence that has been set and the wording after each attribute has become the topic. The body 17 makes the difference freely rewriteable. In addition, the CPU 13 of the host computer 9 is provided with a knowledge word dictionary creation control device 14h, which is used to create the same knowledge-word recognition dictionary as the subject dialogue system shown in FIG. 7; a directory inquiry control device 14b , The computer asks people for the desired directory to specify the context in which they are interlocutors. Moreover, an information item judgment control device 14a is also provided in the CPU13, which can use the knowledgeable word dictionary to perform word decomposition and / or article analysis on the voice input by the user in response to the inquiry of the catalog to determine whether it is used to specify knowledgeability = All information items required by the material; and, a necessary information item consultation control device 14d is used to input information that does not include the information items required to specify the knowledgeable data, which is lacking. Ask users to get all the information items they need. Furthermore, a cache memory recording control device 14 is provided in the CPU 13 so that when the input sound already contains the information required to specify the intellectualized data, the context can be specified according to the directory, and then called from the memory. Those who are designated with knowledgeable data and recorded in the cache memory and made it freely rewriteable; and, a program execution control device 4c, which can make the program performer in accordance with a preset dialogue sequence . In addition, the CPU 13 is provided with a routine control device 14f. When the predetermined attributes and / or levels become the subject of the program in accordance with the dialog sequence, it can call out the intellectualized data for specifying small situations ( Also included are related to this paper standard applicable Chinese National Standard (CNS) A4 specification (21 () > < 297 public director --------- install ------ order ------ line (Please read the precautions on the back before filling out this page) 47367b A7 ------ B7__ printed by the Intellectual Property Bureau of the Ministry of Economic Affairs' Consumer Cooperatives V. Description of the Invention (43) Formula (> Material Library) as a subnormal formula And make the program performers in accordance with the order of the items; and a dialog dialogue control device 14g, which is used to 'revert to the dialog order and make the rest of the programs perform after the end of the item sequence. Next, refer to FIG. 10 for knowledge The construction method of the database is explained in detail. Fig. 10 is a flowchart illustrating the method of constructing the theme dialogue in the third and fourth aspects of the present invention / the method of constructing a knowledgeable database for use. Fig. 11 is a flowchart A table showing several examples of knowledgeable words. Reference No. 10 Series If the market belongs to a variety of topics, industries, and fields (a collection of information related to each topic, industry, and field is referred to as a "situation"), β is a foreign language document 10a, which is translated into the language by a translator 12a. Text, and then send it to the special control device 20. If the data is a general document, it is pre-processed in OCRUb, if the data is sound, it is pre-processed by a well-known method such as voice recognition software 12c and then It is sent to the specialization control system 20. In addition, the right data is various electronic information. The format is unified by the format conversion device 12d, and then sent to the specialization control device 20. The knowledgeable word dictionary is used in the form. For the purpose of monolingual decomposition and / or article parsing, the dialogue with people as input information and various databases are specialized. Moreover, the dictionary of knowledgeable words here stores words classified according to their respective names, Those who use the name as the classification name, the type, size, and available range of the name as its attributes, and the value of the name as the level of the educated word. As in Section 4 When the situation (topic) includes multiple categories, it means that the paper size of each data collection sheet that belongs to a specific situation (topic) is applicable to _ Home Standards (CNS) Public Celebration ____.—__—-. 衣 1T --- ---. ^ (Please read the notes on the back before filling this page) -46 · , 7 丨 0
發明説明(44 ) 然後’將該特定情境之相關資料中所出現之單詞依各 稱加以刀類’並以作為該分類名之名稱與該名稱之類 ’大J及可取得之範圍等條件為其屬性。而,該屬性盘 =屬体之名稱值之等級的組合則成為與各單詞相對應之知 早詞。對於以如上程序作成之知識化單詞,可授與用 ^疋該情境之識別因子。藉此,即可由知識化單詞辭典 中,迅·速檢索屬同一情境之單詞。 「作為分類名之名龜焱Γ头μ 、「…(名稱」係指「季節」、「地名」、「期 斥求」 表不行動之動詞」等「詞類等單詞之 種類+性質瘙拉,Μ: ±< ^ μ龍」。精授與性践特性,該單詞本身及/ 早詞所出現之文章内之其他單詞將產生某種限制。舉 ::之,「人名」之内容或值(等級)中應為文字,而不可 為數字(漢字數字除外)或記號。又,動詞「泊」(或其=二? ’皆意指「住宿」)之主語為「表示人之名詞」 :名敎人名」,而主語與慣用語間則為表示「場 所或地名」之名詞。 另,動詞「;白」(或其語幹「泊」)可與表示「誰 厂 装------、訂------線、 (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 跟誰Description of the Invention (44) Then, 'the words appearing in the relevant information of the specific situation are classified as knives according to each name', and the conditions such as the name of the classification name and the name of the name 'big J and the range that can be obtained are: Its properties. However, the combination of the rank of the attribute disk = the name value of the belonging body becomes an early word corresponding to each word. For knowledgeable words created using the above procedure, ^ 识别 the recognition factor for the situation can be granted. In this way, you can quickly and quickly search for words that belong to the same context from the dictionary of knowledgeable words. "The name of the tortoise 焱 which is the name of the classification," ... (name) means "season", "place name", "period repulsion", "verbs that do not act" and other types of words such as parts of speech + nature, Μ: ± < ^ μ 龙 ". Intensive teaching and sexual practice characteristics, the word itself and / or other words in the article in which the early word appears will have some kind of restriction. For example: the content or value of" person name " (Level) should be text, not numbers (except Chinese characters and numbers) or signs. Also, the subject of the verb "泊" (or == ?? means "accommodation") is "a noun representing a person": "Name, name", and subject and idiomatic terms are nouns representing "place or place name". In addition, the verb "; white" (or its stem "Park") can be used with the meaning "who is factory-installed ------ , Order ------ line, (please read the notes on the back before filling out this page)
J 厂 何時When J Factory
JJ
J 在何處 多少錢 7 /叫」 、!如 付軌」之單詞相連結。此時,舉例言之,「去何處」之 容編須為表示場所之名詞。因此,知識化單詞辭典 之作為分類名之名稱」僅可選擇「表示場所之名叫」 即固有名詞之「地名」與—般名詞中表示場所之單叫, 加以參照以得到業經單詞分解之文章的正確解釋。” 如上所述,知識化單詞係以「作為分類名之名稱」 如何 内 中 而Where is J and how much is it? Such as "pay rail". At this time, for example, the content of "where to go" must be a noun representing a place. Therefore, the name of a dictionary of knowledgeable words as a classification name can only be selected as "the name of the place", that is, the "place name" of the proper noun and the single name of the place in the general noun. Correct interpretation. As mentioned above, the educated word is "in the name of the classification name"
本紙張尺度適用中國國家標準(CNS -47- 47367 b 五、發明説明(45 ) ^該名稱之類型、大小及可取得之範”條件」 且以該名稱之内容作為「等級」而作成者,錢乃記錄. 儲存於知識化單詞辭典30中。 數例之表格。 第η圖即顯不知識化單詞之 特殊化控制裝置20係可使用知識化單詞辭典30,對由 與人對話而輸入之聲音資訊或各種形態之資料、由既存之 貧料庫_中抽出之資料,進行單詞分解及/或文章解析,再以 之作為由屬性及等級所構成之要素單元之集合而加以特殊 化者4其亦可以之作為由設計成可參照知識化單詞辭 典30内之屬性的名稱及等級所構成之要素單元之集合而加 以特殊化。且,其並可大幅節省知識化資料庫利之記憶容 量。 舉例言之,就欲將以下2種文書資料(係藉〇cr自印刷 品讀取者)知識化資料化之,情況加以說明’即,「秋色之北 海道3日遊。函館、小樽、積丹半島。住宿嚮往的公主飯店。 賞楓景點有大沼公園、京極噴泉公園、日光。北海道獨有 之味覺響妥。第一天晚餐是毛蟹御勝。第二天午餐是生魚 片哥司飯。第一天晚餐是公主飯店自助餐。第三天午餐是 蟹肉飯與早魚火銷。母2人一間房。出發日為1〇月9〜12, 16〜19日」,「極致旅遊。乘jAB航空周遊夏威夷3島6日遊1。 出發日與費用另以表列出。全部附餐、觀光行程。和風鐵 板燒·迎賓晚餐加乘船賞夕陽、2島觀光。庭園之島康愛島。 溪谷之島茂伊島。第28回JAB檀香山馬拉松受理報名中」。 第12(a)及(b)圖係顯示已對各文章進行單詞分解後之 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) (請先閱讀背面之注意事項再填寫本頁) .裝. 訂 經濟部智慧財產局員工消費合作社印製 -48- 4736?1j A7 B7 五、發明説明(46 經濟部智慧財產局員工消費合作社印製 狀態者°在第一例中,由於有接續詞「之」及數字「3」後 接有「日」’故特殊化控制裝置20可辨識「秋色」及「北 海道」為應行單詞分解之單位。「秋色」雖可授與「季節 之名但仍須查證是否與記錄於「季節」之屬性中之類 型、大小、其他限制吻合。 又’在「住宿嚮往之公主飯店」一文中,將由表示行 動之動·詞「泊 」產生『主語應為「表示人之名詞或人 名」’且’主語與作為述語之動詞間應為「表示場所之 詞或住所」』之限制。特殊化控制裝置20則可由用以指 主语位置之「是」或「係」並不存在,而判斷其為省略 之文早其久’再就储存於知識化单詞辭典30之名稱r 不場所之名詞」或「固有名詞」中檢索位於表示場所之 詞「在」前之單詞「公主飯店」,以將「公主飯店」抽出 第13(a)及(b)圖係以如上程序作成之知識化資料之 例。在本實施例中,知識化資料係構築以作為由屬性之 % (並未就限制條件加以記錄。)與等級所構成之要素單 之集合者。當然,當知識化資料庫40之容量極大時,即可 形成知識化資料以作為由屬性(亦就限制條件加以記錄、 與等級所構成之要素單元之集合並加以儲存。由於無須 照知識化單詞辭典30,以特殊化控制裝置2〇即可進彳^預 之處理,故處理速度將因此加快。 另,*於該等檢索之應檢索對象將受屬性之名稱及限 ,i、件所局限,故可大幅減少檢索所需之 昇辨識精確度。 门時亦拯 ( CNS ) A4^ ( 210X297^« 名 示 主表 助 名 元 參定 批衣 1訂— — I I線 (請先閱讀背面之注意事項再填苟本頁) -49·This paper size applies the Chinese national standard (CNS-47-47367 b. 5. Description of the invention (45) ^ The type, size and available range of the name "conditions", and the content of the name as the "level", The money is recorded. It is stored in the knowledgeable word dictionary 30. Tables of several examples. Figure η shows the specialization control device 20 for displaying uneducated words. The knowledgeable word dictionary 30 can use the knowledgeable word dictionary 30 for inputting by dialogue with people. Voice information or various forms of data, data extracted from the existing poor library_, word decomposition and / or article analysis, and then specialize it as a set of element units composed of attributes and levels 4 It can also be specialized as a set of element units that are designed to refer to the names and levels of attributes in the knowledgeable word dictionary 30. Moreover, it can greatly save the memory capacity of the knowledgeable database. For example In other words, I want to make the following two kinds of documents and materials (from the readers of printed materials borrowed from 0cr) knowledgeable and informative, and explain the situation ', that is, "Three-day trip to Hokkaido in autumn. Museum, Otaru, Shakotan Peninsula. The princess hotel longing for accommodation. Maple-seeing spots include Onuma Park, Keiji Fountain Park, and Nikko. Hokkaido's unique taste is sound. The first day of dinner is hairy crab mitsuke. The next day is lunch of raw fish. Pian Ge Si Fan. The first day dinner is a buffet at the Princess Hotel. The third day lunch is crab meat rice and early fish. The room for 2 mothers. The departure date is October 9 ~ 12, 16 ~ 19. "Extreme tourism. JAB Airlines travel to Hawaii 3 islands 6 days tour 1. Departure days and fees are listed separately. All meals, sightseeing itinerary. Japanese style teppanyaki, welcome dinner plus boat sunset, 2 island sightseeing .Kangai Island, the island of the garden. Maui Island, the island of the valley. The 28th JAB Honolulu Marathon is currently in the process of registration. China National Standard (CNS) A4 specification (210X297 mm) (Please read the precautions on the back before filling this page). Packing. Order printed by the Intellectual Property Bureau of the Ministry of Economic Affairs Consumer Cooperatives -48- 4736? 1j A7 B7 V. Invention Description (46 Member of Intellectual Property Bureau, Ministry of Economic Affairs Those who have printed the status of the consumer cooperatives. In the first example, because there is a continuation word "zhi" and the number "3" followed by "day", the special control device 20 can recognize "autumn colors" and "Hokkaido" as the response. The unit of word decomposition. Although "autumn" can be given the name of "season", it is still necessary to check whether it is consistent with the type, size, and other restrictions recorded in the attributes of "season." In the text, the restriction that "the subject should be" the name or name of a person "and" the subject and the verb that is the predicate should be "a word or a place of residence" between the subject and the verb that expresses action. Special The control device 20 can be used to indicate that the subject position "is" or "department" does not exist, and it is judged that it is an omitted text long ago, and then stored in the name r of the intellectualized word dictionary 30. Search for the word "princess hotel" in front of the word "before" in the "noun" or "proprietary noun" to extract the "princess hotel" from Figure 13 (a) and (b). Information . In this embodiment, the knowledge-based data is constructed as a set of element lists composed of% (attributes are not recorded) and attributes. Of course, when the capacity of the knowledge database 40 is extremely large, the knowledge data can be formed as a set of element units composed of attributes (also recorded in terms of constraints and levels) and stored. The dictionary 30 can be processed in advance with a special control device 20, so the processing speed will be accelerated accordingly. In addition, the objects to be searched in these searches will be limited by the names and restrictions of the attributes. Therefore, the recognition accuracy required for searching can be greatly reduced. The door is also rescued (CNS) A4 ^ (210X297 ^ «Name shows the main table to help the Yuan Yuan to determine the batch of clothes 1 order-II line (please read the back first (Notes on this page) -49 ·
Claims (1)
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2000/006759 WO2002029633A1 (en) | 2000-09-29 | 2000-09-29 | Topics dialog method and system |
Publications (1)
Publication Number | Publication Date |
---|---|
TW473675B true TW473675B (en) | 2002-01-21 |
Family
ID=11736542
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW089119845A TW473675B (en) | 2000-09-29 | 2000-09-26 | Topics dialog procedures and system |
Country Status (3)
Country | Link |
---|---|
JP (1) | JPWO2002029633A1 (en) |
TW (1) | TW473675B (en) |
WO (1) | WO2002029633A1 (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10541997B2 (en) * | 2016-12-30 | 2020-01-21 | Google Llc | Authentication of packetized audio signals |
WO2019150583A1 (en) | 2018-02-05 | 2019-08-08 | 日本電気株式会社 | Question group extraction method, question group extraction device, and recording medium |
CN112115248B (en) * | 2020-09-27 | 2023-11-14 | 支付宝(杭州)信息技术有限公司 | Method and system for extracting dialogue strategy structure from dialogue corpus |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS62103731A (en) * | 1985-10-30 | 1987-05-14 | Fujitsu Ltd | Retrieval sentence input control system |
-
2000
- 2000-09-26 TW TW089119845A patent/TW473675B/en active
- 2000-09-29 WO PCT/JP2000/006759 patent/WO2002029633A1/en active Application Filing
- 2000-09-29 JP JP2002533135A patent/JPWO2002029633A1/en active Pending
Also Published As
Publication number | Publication date |
---|---|
JPWO2002029633A1 (en) | 2004-02-12 |
WO2002029633A1 (en) | 2002-04-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Mahlberg et al. | CLiC Dickens: Novel uses of concordances for the integration of corpus stylistics and cognitive poetics | |
Joudrey et al. | Introduction to cataloging and classification | |
CN103649953B (en) | method and system for processing a search request | |
CN109086408A (en) | Document creation method, device, electronic equipment and computer-readable medium | |
JP4347226B2 (en) | Information extraction program, recording medium thereof, information extraction apparatus, and information extraction rule creation method | |
Dash et al. | History, features, and typology of language corpora | |
CN108647194B (en) | Information extraction method and device | |
WO2002035369A1 (en) | Data relay system, data relay method, data relay program, information processing method, and information processing program | |
US20090150341A1 (en) | Generation of alternative phrasings for short descriptions | |
Embley et al. | Multilingual ontologies for cross-language information extraction and semantic search | |
Hlava | The taxobook: Principles and practices of building taxonomies, part 2 of a 3-part series | |
Wirawan et al. | Balinese historian chatbot using full-text search and artificial intelligence markup language method | |
Hassan et al. | A Model for brand personality word embedding: Identifying UNESCO world heritage personality categories | |
Lewins | Computer assisted qualitative data analysis software (CAQDAS) | |
Schanding et al. | Shell noun use in English argumentative essays by native speakers of Japanese, Turkish, and English: Frequency and rate of noun-pattern attraction | |
TW473675B (en) | Topics dialog procedures and system | |
JPH0798708A (en) | Document processing system/method | |
JP4057962B2 (en) | Question answering apparatus, question answering method and program | |
KR101072147B1 (en) | Method and system for transforming blog posting to Ontology-based information | |
Chujfi et al. | Machine learning and human cognition combined to enhance knowledge discovery fidelity | |
KR100480345B1 (en) | Method and system for serving language learning/translation using language code | |
Hashim | Syntactic choices and text organization in medical research articles | |
JP2019021337A (en) | Conference support device, conference support device control method, and program | |
KR102340404B1 (en) | Method and apparatus for managing movie recommended items using language units | |
US20240168998A1 (en) | Information processing apparatus, information processing system, information processing method, and non-transitory recording medium |