每年在西班牙舉辦兩次的API盛會,今年於十一月22與23號成功於巴塞隆納mobile world center舉行,以往的主題跨足自然語言處理(NLP)、開放資料(Open Data)等面向,這次盛會則聚焦於聊天機器人的發展以及其在API上的應用與管理。
APIDays is held in Spain twice a year, and it was held successfully in Barcelona on 22th and 23th November. The event topic is relevant to NLP, Open data and it focuses on chatbot for this round.

與會者多半來自南歐地區,開發者佔大多數,少數人為UI設計師與公司主管,而講者則來自美國、德國、瑞典、法國等地區,多達二十位講者的精心分享,讓兩天的議程充實而精采!會議內容由淺入深,議程安排的主題性明確,第一天從API與機器人發展的大方向開始,到機器人發展現況,接著建機器人的實作與設計,進而談到未來發展,第二天則是從機器學習、自然語言處理與自動化的主軸切入,到API設計與機器人管理。讓會眾能輕鬆選擇想聽的場次,會期間皆供應精緻茶點,提供豐富的交流機會,現場當然熱鬧可期!
Attendants mostly came from south Europe, and most of them are developers while few are UI designers and company managers. In terms of speakers, they come from US, Germany, Switzerland, France and Spain. Up to 20 speakers share amazing speeches which makes this APIDays successful and wonderful! The agenda of APIDays were well-organized and clear from the whole picture of chatbot to the current trends, the development and implementation of chatbot and the future of chatbot.

第一天從聊天室在歐洲發展的歷史談起,從1970年起盛行的BBS,接著有UNIX talk
、IRC、ICQ、AIM、Jabber、Google talk、WhatsApp到2010年開始風行的Slack,演變的過程中可以見到幾項趨勢,開發者本身也熱愛聊天,數據統計有高達1800萬的開發者使用聊天室,在聊天室裡分享程式碼,交換經驗,漸漸地導向研發聊天機器人,從命令列下指令的聊天機器人到對話式機器人,至今蛻變成智慧型聊天機器人。有講者提到機器人可能會漸漸取代App,有鑒於機器人的應用越來越廣泛,如何掌握這波潮流將會是未來的研發重點。而另一方面,有講者呼籲在研發過程中,不太可能做出萬能機器人,若是能集中心力研發特定領域,專注目標客戶需求,才能鑽得更深。
The first day started from the history of chat in Europe, such as BBS from 1970 and then UNIX talk, IRC, ICQ, AIM, Jabber, Google talk, WhatsApp and Slack from 2010. From the history, we can understand the trends behind, developers love to chat as well. According to statistics, more than 18 million developers use chat rooms, share their codes and exchange their experiences. Therefore, they started to develop in chatbot. The focus changes from command line chats to conversational bots and now is intelligent chatbots. One speaker said that bot will replace app little by little. In a view of the applications of bot is getting more and more popular, how to seize this opportunity is the main topic for the future. On the other side, some speakers emphasized on the focus of the domain which a company may want to specialized. Don’t do everything in one bot, instead, try to target on specific field. Thus, you can concentrate on it and meet all the needs of your target customers.

比較有趣的是關於UI的討論,在設計聊天機器人過程中,越來越重視使用者的用戶體驗,舉例來說,如何讓使用者願意持續跟機器人對話,在後端設計上如何有效地引導,又或是如何區分新用戶與有經驗的老客戶,在對話的過程中有何種區隔等等。已經有實體機器人在西班牙與法國幾間家樂福實測,將機器人設點於酒櫃前,讓消費者諮詢適合的酒品,經統計發現,第一次使用的消費者感覺非常新鮮,會想試第二次,而一旦消費者發現機器人回答了無新意,總是推薦幾款酒品時,便會感到無趣,使用機器人的意願大幅降低,這凸顯了機器人背後軟體設計的重要性,如何不斷更新以吸引消費者使用。來自法國的新創團隊Citron,成功建立AI聊天機器人,吸引非常多年輕用戶,透過持續追蹤用戶的習慣,針對用戶想問的地方,推薦在巴黎最適合的地點,還提供一些冷門卻實用的資訊,像是週日仍會開門的超市(因為在歐洲大部分週日商店都會休息),該團隊在設想情況上非常周到,也是其脫穎而出的原因。

Another interesting topic is the discussion of user interface. We can figure out that company pay more and more attention to the user experience during the design of a chatbot. For example, how to keep a user chatting with the bot by guiding a user? That might be relevant to the design from the backend. Another example, how to distinguish between a new user and an experienced one? How do you design the conversation to make the segmentation work? There are several robots already set up in some Carrefour markets in France and Spain. They are allocated in front of the wine section to help customers find the most suitable wine. Based on the survey, customers might interact with the robot once, twice but don’t want to continue since the bot only suggest those wines which are repetitive. Customers feel bored with it. This story tells us that how important the software designed in the chabot so as to attract customers all the time. There is one startup from France called Citron, which built an AI chatbot successfully and attract many young users. By tracking a user’ habits and the locations where a user frequently asks, Citron will thus suggest the best place in Paris to that user. Besides, Citron also offers some special but useful information such as those supermarkets which open on Sunday. (Because most of the markets close on Sunday in Europe) Citron think thoroughly for their customers which makes them outstanding.

而API方面,越來越重視圖像化介面,讓開發者外的人員也能輕易上手,如IBM開發的Watson,可用機器學習演算法輕易建立聊天機器人,還能在後端分析所有資料,包含文字、圖片、影片、語音等等。LUIS(Language Understanding Intelligent Service)為免費的語言智能API,可建立在任何資料庫,如Oracle、NoSQL等等,提供創新服務Bot-as-a-Service (BaaS),讓開發者不用再為訓練模型而煩惱,只要有足夠的用戶便能達到不錯的效能。現有的API追求人性化,而如何整合這些API並更有效地開發機器人,將是成功的關鍵。

As for API, more and more attention has been paid to graphical interface of an API, which can help people easily develop and not just for developers. For instance, IBM Watson can help a user build a chatbot very easily by applying machine learning algorithm and also offer the analytics of all kinds of data, such as context, picture, video and voice in the backend dashboard. Another example is LUIS (Language Understanding Intelligent Service), which a free intelligent language API and it can base on any databases, such as Oracle, NoSQL, etc. They offer an innovative service like Bot-as-a-Service (BaaS), and users can create models for their application to better understand intents or entities.

In conclusion, nowadays APIs tend to be more and more user friendly. The key to success is how to integrate these APIs and apply them to develop in chatbot more efficiently.

 

『巴賽隆納APIDays Barcelona由BotPartner 歐洲參與活動協助』

<延伸閱讀>
法國新創團隊Citron http://www.citron.paris/
IBM Watson http://www.ibm.com/watson/
LUIS https://www.luis.ai/

All Posts
×

Almost done…

We just sent you an email. Please click the link in the email to confirm your subscription!

OKSubscriptions powered by Strikingly