跟讀練習: How to Build AI ChatBot with Custom Knowledge Base in 10 mins - 通過影片學習英語口語

選擇練習模式

每種模式都會開啟全屏練習室。

124 句
如果句子過短或過長,請點選 Edit 進行調整。
1
what if you could turn your website
2
and all of those pdf documents manuals guides ebooks
3
and reports into a helpful chatbot imagine asking questions
4
and getting instant answers directly from your website your documents
5
or any ebook well guess what i found a way to
6
do this in less than 10 minutes in this video i'll
7
show you how easy it is to build your own ai
8
chatbot with custom knowledge base without any coding skills you can do this for your business
9
or other businesses
10
and start making money by the end of this tutorial video
11
not only you'll be able to create a chat bot
12
and train your ai based on your own data using pdf html txt csv word documents
13
and powerpoint presentation files but also integrate the chatbot into your website facebook messenger Slack, Discord, or Telegram.
14
In this video, you'll learn how to create a data store agent, how to add unstructured data to your agent's data store,
15
how to enable voice and chat experiences for your agent, how to test agent and simulate customer questions, how to view conversation history and analytics for your agent,
16
all of that in about 10 minutes.
17
Without any further ado, Let's get started.
18
Alright, so before we start, let me show you the document that we'll be using as the data source.
19
This is budget of the US government and it's 188 pages long.
20
So to get started, we'll go to cloud .gibble .com and log in with your account and click on console.
21
So, from the console, click Search for Agent Builder.
22
Click on Agent Builder.
23
Before creating a new app, make sure that the Vertex AI API and the Dialogflow APIs are enabled.
24
If they're not enabled, you'll get a prompt and then you can enable it on the screen.
25
Select Chat and provide the company name.
26
Provide the agent name
27
Click continue
28
We'll create a datastore This is where we can create a datastore Click on create datastore
29
You get three options here You can provide a website URL
30
And it will automatically crawl website content From a list of domains
31
that you defined You can provide the data from your storage bucket, or you can import data manually by calling the API.
32
I'll select the cloud storage, and you can import a folder or you can import a file.
33
I'll select File and hit Browse.
34
I created this bucket earlier where I uploaded that PDF document that I showed you.
35
I've already uploaded it.
36
But for your case, you can upload any document to your cloud storage, and you'll be able to select it from here.
37
Click Select.
38
So PDF, HTML, text files, CSV, Word document, and presentation files fall under the unstructured documents.
39
Now, you can use that as the data source, or you can use JSON -L for unstructured documents with metadata information.
40
You can click on this link for more information, or you can use CSV file for structured frequently asked data.
41
This is only for chat.
42
But for our example, we're going to be using the unstructured document.
43
Click on continue.
44
I'm going to name the data store US budget 2025 -25 and click on create.
45
It's going to create a data store.
46
Okay, so it created a data store.
47
I will select the data store and click on create.
48
okay
49
so the data store has been created over here I can
50
click on over on this link to see more information on this screen I can see the activity
51
and I need to make sure that data has been imported successfully without any errors.
52
And we are using the digital parser.
53
Now we can hit preview to see the preview.
54
Now we'll test our agent.
55
Testing is an important part of working with conversational agents to find the bugs, identify limitations and simulate a customer experience.
56
You can start an interactive session with your chatbot
57
So to see how it responds to various questions that a customer might ask, click on the test agent.
58
Link at the top right -hand corner.
59
This is where we can test the chatbot.
60
We can write hello.
61
And we have more control over estimating the response of the agent from the screen.
62
We can enable logging by going here and checking this enable conversational history.
63
By checking this box, we will get the conversation history.
64
Let's suppose we put this on our website and people interact with our chatbot.
65
And if you want to be able to see those interactions, you need to check this box and hit save.
66
We'll click on manage, click on integration.
67
Let me show you a very cool thing.
68
You can add a phone gateway to your bot, which will make use of speech -to -text and text -to -speech capabilities in the Google Cloud.
69
You can hit on this manage button to enable that feature
70
but for now I'm happy with the text chatbot
71
and text based I can click on connect we can enable this unauthenticated API
72
which means that will allow people without authentication to interact with the chatbot
73
and I'll click on enable unauthenticated API this is the code
74
that you can use to put on your website
75
and you'll get the chatbot on your website
76
so to see how it would look like click on try it now
77
and you get this um you get this icon at the bottom uh right hand corner
78
so you click on this and this is the chat box
79
that you you can put on your website what are president's top three hour credits for year 2025 Bye.
80
The chatbot read the document and responded with prison -stopped economic parodies, lowering costs for American families.
81
The administration is working to bring down the prescription drug costs, health insurance premiums, utility bills, costs for everyday goods, all while taking on junk fees that some allies,
82
banks, and other companies used to rip off Americans it also is grounded the data
83
and if i click on this link it will show me exactly where it found
84
that information from and as you can see it's on page nine
85
and this is where the chatbot found
86
that information is lowering costs for hardworking americans the administration is
87
working to bring down prescription drug costs health insurance we are grounding it with our own data source
88
So let me ask it another question.
89
Let me ask, what has the president invested in?
90
The president has invested in America and the American people by signing into law historic legislation, such as the American Rescue Plan,
91
the Bipartisan Infrastructure Law, the CHIPS, the Science Act, and the Inflation Reduction Act.
92
Let me ask you another question like,
93
what are some of the accomplishments of the Biden administration?
94
The Biden administration has made historic investments in mental health care, including nearly $1 billion to support the 988 Suicide and Crisis Lifeline.
95
The administration is also committed to investing in training more mental health professionals and has expanded certified community behavioral health clinics nationwide.
96
When testing your virtual agent, if you find that your bot's answers do not meet your expectations,
97
you can always add more URLs to the data source to improve how your bot handles questions and answers.
98
You can add more URLs, you can add more documents.
99
Other than testing many different paths the customer might make, how else can you determine where the problem and friction points are in your virtual agent?
100
Conversation history and analytics help.
101
So we enabled, I showed you where you can enable that conversation history under the agents.
102
I already enabled that.
103
And to get to that conversation history, you can click on conversation history and this is where you're going to see the conversation history.
104
And I can click on any of these
105
and it will show me my interaction with the chat bot
106
I've asked it's going to show the history of my conversation
107
and once you put that on your website
108
and you would be able to see the interaction of your
109
chat bot with your visitors website visitors congratulations using the verdicts
110
ai conversation in dialogue flow cx you have successfully created a
111
data store agent your virtual agent can answer hundreds of different questions about your website or
112
any document pdf manual guides ebooks and reports etc
113
that you may have you don't have to go through the
114
manual process of creating a large number of intense training phrases response messages.
115
We literally created this in less than 10 minutes.
116
this video was requested in the comment section by a few of my subscribers i read all of the comments
117
and i'd love to hear your thoughts feel free to share your ideas for upcoming videos
118
if you don't have any questions
119
or feedback you can support my work by just typing coding
120
money in the comment section it really helps with the youtube algorithms
121
if you found this video useful a thumbs up would be
122
highly appreciated stay tuned for exciting new videos to ensure you don't miss out please subscribe
123
and hit the notification bell icon you may also want to check out the video
124
that pops up on the screen somewhere until next time thank you for watching

下載應用

流利口語所需的一切

AI 發音評分逐句打分
音標練習掌握每個音
詞彙學習建立你的詞庫
詞彙遊戲邊玩邊學
FOR YOU

推薦影片

本課的詞彙與口語要點

這節口語課以影片“How to Build AI ChatBot with Custom Knowledge Base in 10 mins”為素材。 影片中反覆出現的詞有:click, agent, document, website, chatbot。 這段影片共有 124 個句子、1542 個單詞可供跟讀。 講話部分時長為 10:45。 說話人語速平穩,每分鐘約 143 個詞,很適合跟讀。 82% 的單詞屬於英語最常用的 3,000 詞,其餘的詞建議在練習前先學一下。

影片中的重點詞彙

影片中 15 個值得學習的單詞,附發音和釋義:

單詞發音釋義
select 動詞/sɪˈlɛkt/選 /選, 挑選 /挑選
cloud 名詞/ˈklaʊ̯d/йүн, 雲 /雲
import 名詞/ˈɪm.pɔɹt/進口貨 /進口貨, 舶來品 /舶來品
invest 動詞/ɪnˈvɛst/耗費 /耗費, 投資 /投資
manual 名詞/ˈmæn.j(u)əl/手冊 /手冊
interaction 名詞/ˌɪn.təˈɹæk.ʃən/相互作用
upload 動詞/ˈʌpˌloʊd/上載 /上載, 上傳 /上傳
icon 名詞/ˈaɪ.kɑn/聖像 /聖像
bucket 名詞/ˈbʌkɪt/水桶, 吊桶
interact 動詞/ɪn.təˈɹækt/互動 /互動
visitor 名詞/ˈvɪzɪtɚ/訪客 /訪客, 客人
prescription 名詞/pɹəˈskɹɪp.ʃən/藥方 /藥方, 處方 /處方
analytics 名詞/ˌæn.əˈlɪt.ɪks/分析學 /分析學
everyday 形容詞/ˈɛvɹiˌdeɪ/日常的
custom 名詞/ˈkʌstəm/習慣 /習慣, 習俗 /習俗

影片中出現的短語動詞

單詞釋義
check out 動詞退房
go through 動詞通過 /通過
miss out 動詞錯過 /錯過, 放過 /放過

影片中的語法

說話人最常用的結構,並附上影片中的原話:

結構影片中的用法
條件句 if + 從句,will/would + 動詞 — 條件及其結果If they're not enabled, you'll get
現在完成時 have/has + 過去分詞 — 過去發生但與現在仍有關聯的事I've already uploaded · has been created · has been imported
被動語態 be + 過去分詞 — 強調發生了什麼,而不是誰做的are enabled · has been created · has been imported

需要注意的發音

說話人用了 25 次縮略和弱讀形式,例如 we'll, you'll, I'll。請按聽到的簡短形式來說。

  • “th” 音: algorithm /ˈælɡəɹɪðm̩/, thumb /ˈθʌm/, authentication /ɔːˌθɛntɪˈkeɪʃn̩/
  • “sh” 和 “zh” 音: interaction /ˌɪn.təˈɹæk.ʃən/, presentation /ˌpɹɛzn̩ˈteɪʃn̩/, prescription /pɹəˈskɹɪp.ʃən/, congratulations /kənˌɡɹæt͡ʃəˈleɪʃ(ə)nz/, integration /ˌɪn.tɪˈɡɹeɪ.ʃən/
  • 長單詞——注意重音位置: interaction /ˌɪn.təˈɹæk.ʃən/, successfully /səkˈsɛs.fə.li/, analytics /ˌæn.əˈlɪt.ɪks/, congratulations /kənˌɡɹæt͡ʃəˈleɪʃ(ə)nz/, utility /juːˈtɪlɪti/

中文母語者容易讀錯的音:

  • /v/ — 不要讀成 /w/,上齒輕觸下唇: invest /ɪnˈvɛst/, virtual /ˈvɜɹt͡ʃuəl/, visitor /ˈvɪzɪtɚ/, preview /ˈpɹiːvjuː/, everyday /ˈɛvɹiˌdeɪ/
  • 詞尾子音 — 要讀清楚,後面不要加母音: select /sɪˈlɛkt/, cloud /ˈklaʊ̯d/, import /ˈɪm.pɔɹt/, invest /ɪnˈvɛst/, upload /ˈʌpˌloʊd/
  • /r/ 和 /l/ 的區別: virtual /ˈvɜɹt͡ʃuəl/, congratulations /kənˌɡɹæt͡ʃəˈleɪʃ(ə)nz/, algorithm /ˈælɡəɹɪðm̩/, behavioral /bɪˈheɪvjəɹəl/, tutorial /ˌtjuːˈtɔːɹɪəl/

如何用這段影片練習

  1. 先完整聽一遍影片,不要開口,記下不認識的單詞。
  2. 用正常速度逐句跟讀,每句重複到你的節奏與說話人一致為止。
  3. 錄下自己的聲音並與原聲對比,特別注意 select, cloud, import 這類單詞。

什麼是跟讀法?

跟讀法 (Shadowing) 是一種有科學依據的語言學習技巧,最初開發用於專業口譯員的培訓,並由多語言者Alexander Arguelles博士普及。這個方法簡單而強大:您在聽英語母語原聲的同時立即大聲重複——就像是一個延遲1-2秒緊跟說話者的影子。與被動聽力或語法練習不同,跟讀法強迫您的大腦和口腔肌肉同時處理並模模擬實的講話模式。研究表明它能顯著提高發音準確性,語調,節奏,連讀,聽力理解和口語流利度——使其成為雅思口語備考和真實英語交流最有效的方法之一。

影子跟讀法: 閱讀完整分步指南 →