Posted by Andrew Zaldivar, Developer Advocate, Google AI
A few months ago, we announced our AI Principles, a set of commitments we are upholding to guide our work in artificial intelligence (AI) going forward. Along with our AI Principles, we shared a set of recommended practices to help the larger community design and build responsible AI systems.
In particular, one of our AI Principles speaks to the importance of recognizing that AI algorithms and datasets are the product of the environment—and, as such, we need to be conscious of any potential unfair outcomes generated by an AI system and the risk it poses across cultures and societies. A recommended practice here for practitioners is to understand the limitations of their algorithm and datasets—but this is a problem that is far from solved.
To help practitioners take on the challenge of building fairer and more inclusive AI systems, we developed a short, self-study training module on fairness in machine learning. This new module is part of our Machine Learning Crash Course, which we highly recommend taking first—unless you know machine learning really well, in which case you can jump right into the Fairness module.
The Fairness module features a hands-on technical exercise. This exercise demonstrates how you can use tools and techniques that may already exist in your development stack (such as Facets Dive, Seaborn, pandas, scikit-learn and TensorFlow Estimators to name a few) to explore and discover ways to make your machine learning system fairer and more inclusive. We created our exercise in a Colaboratory notebook, which you are more than welcome to use, modify and distribute for your own purposes.
From exploring datasets to analyzing model performance, it's really easy to forget to make time for responsible reflection when building an AI system. So rather than having you run every code cell in sequential order without pause, we added what we call FairAware tasks throughout the exercise. FairAware tasks help you zoom in and out of the problem space. That way, you can remind yourself of the big picture: finding the undesirable biases that could disproportionately affect model performance across groups. We hope a process like FairAware will become part of your workflow, helping you find opportunities for inclusion.
FairAware task guiding practitioner to compare performances across gender.
The Fairness module was created to provide you with enough of an understanding to get started in addressing fairness and inclusion in AI. Keep an eye on this space for future work as this is only the beginning.
If you wish to learn more from our other examples, check out the Fairness section of our Responsible AI Practices guide. There, you will find a full set of Google recommendations and resources. From our latest research proposal on reporting model performance with fairness and inclusion considerations, to our recently launched diagnostic tool that lets anyone investigate trained models for fairness, our resource guide highlights many areas of research and development in fairness.
Let us know what your thoughts are on our Fairness module. If you have any specific comments on the notebook exercise itself, then feel free to leave a comment on our GitHub repo.
On behalf of many contributors and supporters,
Andrew Zaldivar – Developer Advocate, Google AI
Posted by Wesley Chun (@wescpy), Developer Advocate, Google Cloud
Google Cloud Next '18 is only a few days away, and this year, there are over 500 sessions covering all aspects of cloud computing, from G Suite to the Google Cloud Platform. This is your chance to learn first-hand how to build custom solutions in G Suite alongside other developers from Independent Software Vendors (ISVs), systems integrators (SIs), and industry enterprises.
G Suite's intelligent productivity apps are secure, smart, and simple to use, so why not integrate your apps with them? If you're planning to attend the event and are wondering which sessions you should check out, here are some sessions to consider:
I look forward to meeting you in person at Next '18. In the meantime, check out the entire session schedule to find out everything it has to offer. Don't forget to swing by our "Meet the Experts" office hours (Tue-Thu), G Suite "Collaboration & Productivity" showcase demos (Tue-Thu), the G Suite Birds-of-a-Feather meetup (Wed), and the Google Apps Script & G Suite Add-ons meetup (just after the BoF on Wed). I'm excited at how we can use "all the tech" to change the world. See you soon!
Posted by Wesley Chun (@wescpy), Developer Advocate, G Suite
While most chatbots respond to user requests in a synchronous way, there are scenarios when bots don't perform actions based on an explicit user request, such as for alerts or notifications. In today's DevByte video, I'm going to show you how to send messages asynchronously to rooms or direct messages (DMs) in Hangouts Chat, the team collaboration and communication tool in G Suite.
What comes to mind when you think of a bot in a chat room? Perhaps a user wants the last quarter's European sales numbers, or maybe, they want to look up local weather or the next movie showtime. Assuming there's a bot for whatever the request is, a user will either send a direct message (DM) to that bot or @mention the bot from within a chat room. The bot then fields the request (sent to it by the Hangouts Chat service), performs any necessary magic, and responds back to the user in that "space," the generic nomenclature for a room or DM.
Our previous DevByte video for the Hangouts Chat bot framework shows developers what bots and the framework are all about as well as how to build one of these types of bots, in both Python and JavaScript. However, recognize that these bots are responding synchronously to a user request. This doesn't suffice when users want to be notified when a long-running background job has completed, when a late bus or train will be arriving soon, or when one of their servers has just gone down. Recognize that such alerts can come from a bot but also perhaps a monitoring application. In the latest episode of the G Suite Dev Show, learn how to integrate this functionality in either type of application.
From the video, you can see that alerts and notifications are "out-of-band" messages, meaning they can come in at any time. The Hangouts Chat bot framework provides several ways to send asynchronous messages to a room or DM, generically referred to as a "space." The first is the HTTP-based REST API. The other way is using what are known as "incoming webhooks."
The REST API is used by bots to send messages into a space. Since a bot will never be a human user, a Google service account is required. Once you create a service account for your Hangouts Chat bot in the developers console, you can download its credentials needed to communicate with the API. Below is a short Python sample snippet that uses the API to send a message asynchronously to a space.
from apiclient import discovery from httplib2 import Http from oauth2client.service_account import ServiceAccountCredentials SCOPES = 'https://www.googleapis.com/auth/chat.bot' creds = ServiceAccountCredentials.from_json_keyfile_name( 'svc_acct.json', SCOPES) CHAT = discovery.build('chat', 'v1', http=creds.authorize(Http())) room = 'spaces/<ROOM-or-DM>' message = {'text': 'Hello world!'} CHAT.spaces().messages().create(parent=room, body=message).execute()
The alternative to using the API with service accounts is the concept of incoming webhooks. Webhooks are a quick and easy way to send messages into any room or DM without configuring a full bot, i.e., monitoring apps. Webhooks also allow you to integrate your custom workflows, such as when a new customer is added to the corporate CRM (customer relationship management system), as well as others mentioned above. Below is a Python snippet that uses an incoming webhook to communicate into a space asynchronously.
import requests import json URL = 'https://chat.googleapis.com/...&thread;_key=T12345' message = {'text': 'Hello world!'} requests.post(URL, data=json.dumps(message))
Since incoming webhooks are merely endpoints you HTTP POST to, you can even use curl to send a message to a Hangouts Chat space from the command-line:
curl
curl \ -X POST \ -H 'Content-Type: application/json' \ 'https://chat.googleapis.com/...&thread;_key=T12345' \ -d '{"text": "Hello!"}'
To get started, take a look at the Hangouts Chat developer documentation, especially the specific pages linked to above. We hope this video helps you take your bot development skills to the next level by showing you how to send messages to the Hangouts Chat service asynchronously.
We recently introduced Hangouts Chat to general availability. This next-generation messaging platform gives G Suite users a new place to communicate and to collaborate in teams. It features archive & search, tighter G Suite integration, and the ability to create separate, threaded chat rooms. The key new feature for developers is a bot framework and API. Whether it's to automate common tasks, query for information, or perform other heavy-lifting, bots can really transform the way we work.
In addition to plain text replies, Hangouts Chat can also display bot responses with richer user interfaces (UIs) called cards which can render header information, structured data, images, links, buttons, etc. Furthermore, users can interact with these components, potentially updating the displayed information. In this latest episode of the G Suite Dev Show, developers learn how to create a bot that features an updating interactive card.
As you can see in the video, the most important thing when bots receive a message is to determine the event type and take the appropriate action. For example, a bot will perform any desired "paperwork" when it is added to or removed from a room or direct message (DM), generically referred to as a "space" in the vernacular.
Receiving an ordinary message sent by users is the most likely scenario; most bots do "their thing" here in serving the request. The last event type occurs when a user clicks on an interactive card. Similar to receiving a standard message, a bot performs its requisite work, including possibly updating the card itself. Below is some pseudocode summarizing these four event types and represents what a bot would likely do depending on the event type:
function processEvent(req, rsp) { var event = req.body; // event type received var message; // JSON response message if (event.type == 'REMOVED_FROM_SPACE') { // no response as bot removed from room return; } else if (event.type == 'ADDED_TO_SPACE') { // bot added to room; send welcome message message = {text: 'Thanks for adding me!'}; } else if (event.type == 'MESSAGE') { // message received during normal operation message = responseForMsg(event.message.text); } else if (event.type == 'CARD_CLICKED') { // user-click on card UI var action = event.action; message = responseForClick( action.actionMethodName, action.parameters); } rsp.send(message); };
The bot pseudocode as well as the bot featured in the video respond synchronously. Bots performing more time-consuming operations or those issuing out-of-band notifications, can send messages to spaces in an asynchronous way. This includes messages such as job-completed notifications, alerts if a server goes down, and pings to the Sales team when a new lead is added to the CRM (Customer Relationship Management) system.
Hangouts Chat supports more than JavaScript or Python and Google Apps Script or Google App Engine. While using JavaScript running on Apps Script is one of the quickest and simplest ways to get a bot online within your organization, it can easily be ported to Node.js for a wider variety of hosting options. Similarly, App Engine allows for more scalability and supports additional languages (Java, PHP, Go, and more) beyond Python. The bot can also be ported to Flask for more hosting options. One key takeaway is the flexibility of the platform: developers can use any language, any stack, or any cloud to create and host their bot implementations. Bots only need to be able to accept HTTP POST requests coming from the Hangouts Chat service to function.
At Google I/O 2018 last week, the Hangouts Chat team leads and I delivered a longer, higher-level overview of the bot framework. This comprehensive tour of the framework includes numerous live demos of sample bots as well as in a variety of languages and platforms. Check out our ~40-minute session below.
To help you get started, check out the bot framework launch post. Also take a look at this post for a deeper dive into the Python App Engine version of the vote bot featured in the video. To learn more about developing bots for Hangouts Chat, review the concepts guides as well as the "how to" for creating bots. You can build bots for your organization, your customers, or for the world. We look forward to all the exciting bots you're going to build!
Posted by Wesley Chun (@wescpy), Developer Advocate, Google Apps
At Google I/O 2016, we launched a new Google Sheets API—click here to watch the entire announcement. The updated API includes many new features that weren’t available in previous versions, including access to functionality found in the Sheets desktop and mobile user interfaces. My latest DevByte video shows developers how to get data into and out of a Google Sheet programmatically, walking through a simple script that reads rows out of a relational database and transferring the data to a brand new Google Sheet.
Let’s take a sneak peek of the code covered in the video. Assuming that SHEETS has been established as the API service endpoint, SHEET_ID is the ID of the Sheet to write to, and data is an array with all the database rows, this is the only call developers need to make to write that raw data into the Sheet:
SHEETS
SHEET_ID
data
SHEETS.spreadsheets().values().update(spreadsheetId=SHEET_ID, range='A1', body=data, valueInputOption='RAW').execute()
rows = SHEETS.spreadsheets().values().get(spreadsheetId=SHEET_ID, range='Sheet1').execute().get('values', []) for row in rows: print(row)
If you’re ready to get started, take a look at the Python or other quickstarts in a variety of languages before checking out the DevByte. If you want a deeper dive into the code covered in the video, check out the post at my Python blog. Once you get going with the API, one of the challenges developers face is in constructing the JSON payload to send in API calls—the common operations samples can really help you with this. Finally, if you’re ready to get going with a meatier example, check out our JavaScript codelab where you’ll write a sample Node.js app that manages customer orders for a toy company, the database of which is used in this DevByte, preparing you for the codelab.
We hope all these resources help developers create amazing applications and awesome tools with the new Google Sheets API! Please subscribe to our channel, give us your feedback below, and tell us what topics you would like to see in future episodes!
Posted by Josh Gordon, Developer Advocate
To help you get started building applications with machine learning, we’re excited to launch a new developer show, Machine Learning: Recipes for New Developers. In the first few episodes, we’ll teach you the ropes of machine learning without requiring any major prerequisites (like calculus). As the series progresses, we’ll walk you from “Hello World” to solving some real world problems.
Episodes will generally publish bi-weekly, and be only about 5-10 minutes in length to keep the material lightweight. Occasionally, we’ll have guests on the show who work with machine learning on different teams around Google.
Ep #1: Hello World.
Also: Coffee with a Googler came to NYC! Laurence and Josh talk about the importance of machine learning for developers, and reducing barriers to machine learning education. Check out the video!
people
userID
import gdata.analytics.clientAPP_NAME = 'goal_names_demo'my_client = gdata.analytics.client.AnalyticsClient(source=APP_NAME)# Authorizemy_client.client_login( INSERT_USER_NAME, INSERT_PASSWORD, APP_NAME, service='analytics')# Make a query.query = gdata.analytics.client.GoalQuery( acct_id='INSERT_ACCOUNT_ID', web_prop_id='INSERT_WEB_PROP_ID', profile_id='INSERT_PROFILE_ID')# Get and print results.results = my_client.GetManagementFeed(query)for entry in results.entry: print 'Goal number = %s' % entry.goal.number print 'Goal name = %s' % entry.goal.name print 'Goal value = %s' % entry.goal.value
I've always hoped that I could release Mondrian as open source, but it was not to be: due to its popularity inside Google, it became more and more tied to proprietary Google infrastructure like Bigtable, and it remained limited to Perforce, the commercial revision control system most used at Google.What I'm announcing now is the next best thing: an code review tool for use with Subversion, inspired by Mondrian and (soon to be) released as open source. Some of the code is even directly derived from Mondrian. Most of the code is new though, written using Django and running on Google App Engine.I'm inviting the Python developer community to try out the tool on the web for code reviews. I've added a few code reviews already, but I'm hoping that more developers will upload at least one patch for review and invite a reviewer to try it out.
