One of Agoloa��s main areas of focus is text summarization. Summarization is the process of taking a mess of unstructured information and distilling it into something consumable. Easier said than done. This formulation of the problem oversimplifies many considerations that a summarization system has to take into account. Implementing a system comes with choices, and with choices come tradeoffs.

This post lays out some of the most important choices. It also gives an overview of summarization in general. Each of the following sections discusses some of the insights we have gained and the approaches we take to tackle these problems at Agolo.

Abstractive vs. Extractive

When a human is given a corpus of text to summarize, they might rewrite the main points in their own words. This is how many professional human-generated summaries like Cliffs Notes operate. Ita��s called abstractive summarization.

Rewriting in different words requires many high-level skills that only human experts have. Simulating these skills requires coherent natural language generation, an extensive and deep knowledge of context, and adequate modeling of the readera��s mind. The state of the art is not yet up to par, so many automatic summarization systems opt for a technique called extractive summarization.

Extractive summaries are excerpts taken directly from the input documents and presented in a readable way. The summary does not contain any rephrasing of the ideas presented in the original text.

Some challenges in picking these sentences include:

• processing the input text to make it presentable in the summary
• determining which sentences are the most salient
• making the summary cohesive and readable
• minimizing the number of references to ideas and entities not mentioned in the summary

Agolo falls somewhere between an extractive and abstractive summarization service. Our system starts by picking the most salient sentences as a basis for a summary, and then takes into account an extensive knowledge graph andthe discourse to provide context.

Single-document vs. Multi-document

Does the summarizer produce one summary per document, or does it distill multiple documents into a single summary?

When summarizing a single document, the summarization system can rely on a cohesive piece of text with very little repetition of facts. The author of a document would not reveal the same information more than once.

However, if the system summarizes multiple related documents, then it must ensure that the summary doesna��t contain repeated or conflicting information. The research literature sometimes formulates this as an optimization problem. An ideal multi-document summarizer maximizes the important information included in the summary while minimizing repetition.

How do you pick which documents to summarize together in the first place? In a large dataset, how do you know which documents are related and should be summarized together? Agolo solves this by clustering documents together into logical stories before summarizing them. The quality of our clusters ensure that the documents are about the same topic or event.

Indicative vs. Informative

The purpose of the summary is closely tied to its intended audience and their goals.

If the readera��s goal is to gain a cursory understanding of a new or large topic, then the summary needs to be indicative. It should give an overview of the content rather than dissect every aspect. This use case is for people like analysts or journalists who need to keep up on fast-moving news or explore unknown topics. They can choose to dive deeper into the text if they feel the need. But most of the time, the main points of a text is good enough for them.

Decision-makers, on the other hand, need detailed breakdowns of text. In this case, the summary should be informative. It should analyze in detail every topic covered in the text. The summary should almost be a replacement for the original text.

Depending on the use case, input documents, and other factors, Agoloa��s summarization system provides both indicative and informative summaries. We provide personalized summaries tailored to each clienta��s needs. Personalization at this level is based on a close relationship with our users to understand their goals and concerns.

Generic vs. Query-based vs. Domain-specific

How users consume the summaries greatly influences summarization.

A summarization system with whata��s called a generic trigger will find the most important topics in a given input text and summarize it without further guidance. For example, a system could produce a summary of the most important, real-time information about a hurricane as news articles are being published.

A generic trigger for summarization is useful in cases where the user does not yet know the contents of the text to be summarized. This is a challenging use case because it cannot rely on human intervention. The summary needs to present the most important topics, which might serendipitously provide the user with new knowledge. Identifying topics and determining which ones the author considers important are difficult challenges.

On the contrary, a query-based summary starts with a topic or question. A query-based summarization system may be given a large corpus of research papers to summarize the effects of a specific chemical compound on the environment. The system first needs to find only the papers that mention that chemical compound, identify the sections that mention its effects, and then summarize the salient points. The system needs to be able to discern the thread of each topic weaving through the text, and then provide a concise version of that discussion.

A summarizer could also use the texta��s domain. For instance, a blog post about a recipe will use different jargon than a blog post about the acquisition of a tech company. A summary meant for experts in a domain can leverage taxonomies or a knowledge graph to take advantage of jargon.

For example, a financial analyst would not need the summary to define the word a�?acquisition.a�? This sense of the word means something different than the same word in an article about a museum acquiring paintings. Taking domain into account, the summarizer can expect sentences about mergers to be related to sentences about acquisitions.

Agoloa��s summarization system is designed to handle text in a variety of domains depending on the client and features derived from the input text.

Genre and OtherA�Factors

A summarization system works best when it knows in advance what kind of text it will encounter. Leta��s call these formal properties.

The length of the input text heavily impacts the sort of approaches a summarization system can take. A typical news article can be summarized with conventional extractive summarization techniques. On the other hand, a 20-page report or a chapter of a book can only be summarized with the help of more advanced approaches like hierarchical clustering or discourse analysis.

Another consideration is the original form of the text: spoken vs. written.Spoken language is different from written language. Transcripts of spoken language are more likely to contain ungrammatical utterances with lots of repetitions.

A related consideration is the genre of the input documents. The discourse structure of an earnings call transcript is different from that of a 10-K filing. Dialog analysis is relevant to genres like chat logs, emails, and customer service phone call transcripts, which record conversations among several people. Dialog contains more topic shifts, interruptions, and anaphora than other genres.

A good summarization system can leverage genre information and other formal properties to produce useful summaries. Agoloa��s summarizer takes these factors into account at various points in the summarization process.

In Summary

Summarization is more complicated than it seems.

When a machine generates a summary, it needs to take into consideration:

• the level of the output summarya��s abstractiveness
• the number of input documents per summary
• the purpose of the summary
• the type of trigger for the summary
• genre and other formal properties of the input text

Wea��ve proven with Agolo ita��s possible to fall somewhere between these hard choices to best satisfy usersa�� needs. Our clients have a wide range of use cases that span all of these considerations and more. Our summarization system is designed to be flexible and extensible to suit these needs. This is one of the most challenging aspects of summarization at Agolo.

This post is only a high-level overview of these considerations. For further reading and a more detailed typology, I recommend the following resources:

• a�?Automatic Summarizationa�? by Nenkova and McKeown
• a�?Introduction to the Special Issue on Summarizationa�? by Radev, Hovy, and McKeown
• a�?Automated Text Summarization and the SUMMARIST Systema�? by Hovy and Yew-Lin

Socks, coasters, mints, hand sanitizer, candy, usb drives, flashlights, backpacks, notebooks, pens, water bottles, tote, phone chargers, t-shirts, keychains, moleskins, phone cases, paper weights, sunglasses, bluetooth speakers, baseball hatsa��.

Those are just some of the items our CEO has collected at the 300+ conferences hea��s attended over the years.

How many socks, coasters, mints, hand sanitizers, usb drives, backpacksa�� have you collected from conferences? Do you still have them? Do you remember who gave them to you?

Can we really prove that typical conference swag impacts brand awareness and affinity? Can we really track that sale down to the tin of mints?

With over five hundred conferences under our teama��s belt, we cana��t say it does.

The NewA�Standard

Brands that are pushing past the plastic and engaging event attendees in experiences are creating lasting impressions in new ways. Tinder was thinking about an engaging experiences when they launched their Tinder VR gag at CES this year. Bumble upped their game at SXSW with the Bumble bus this year. Top brands and businesses are creating new standards for event marketing teams by utilizing experiences and engagement.

So how do brands move beyond conference swag into event experiences?

Top brands are creating digital experiences that attract potential users like Shopifya��s logo maker or accidentally creating an entire business because you had extra stock photos lying around.

In terms of inbound marketing methodology, the people utilizing these tools are top of funnel.

Conferences are the same top of funnel events for potential customers, just in the real world. That means brands need to attract customers with useful tools, resources, and content so when theya��re ready to make a purchasing decision, they have developed brand awareness and affinity.

Transitioning to event experiences over typical conference swag provides brands the opportunity to stand out, make lasting impressions, and create a new standard in event marketing. By taking the inbound methodology into the conference and event space, businesses can begin to stand out from the noise of socks, coasters, mints, hand sanitizer, candy, usb drives, flashlightsa��.

Think about Experiences

Event experiences like Tindera��s VR hoax or Bumblea��s bus can be time intensive and potentially expensive experiences, but they dona��t have to be. Here are a few ways brands can provide valuable event experiences instead of handing out traditional conference swag.

Snapchat Filters

Snapchat isna��t just for 13 year olds, in fact 38% of Snapchat users are between 25 and 54 and with the average user opening the app 18 times a day, ita��s a channel brands can utilize even more at events.

Seriouslya��This is probably one of the most underutilized mirco-moments at conferences. By creating a Snapchat geofilter that is both event and brand specific, companies can begin to capture and influence these personal moments.

Brands can use pre-made templates or hire a Snapchat artist like CyreneQ to create filters which can then be geo-targeted down to a specific building. We quickly went through the steps targeted the entire block of the Moscone Center in San Francisco for the four days of Marketing Nation Summit and the ad would cost $1,035.81. For reference, thata��s about 98.6485714 pairs of custom socks.

Yes, these snaps might disappear in Snaplandia, but the micro-moment that a user has of taking a selfie at this huge event, surrounded by thousands of peers doesna��t. Plus, they can always download the photo and post to Instagram, Facebook, and Twitter to make sure all their friends know theya��re at the event.

Going Live

With Facebook and Instagram Live, wea��re pretty certain authentic live video content is here to stay (and dominate) in 2017. Brands who can figure out ways to utilize live video in a way that excites both those actively participating and those watching are able to reach multiple audiences in one unique fell swoop.

Creating live content is all about representing your brand authentically. This means that big budgets are not required to create fun, entertaining, and unique live content opportunities. Set up a Q&A with attendees, interview a speaker or two live, or walk around the conference floor so people who couldna��t make it get to partake.

Product Giveaways

Dona��t pull a full on Oprah Winfrey when planning a product giveaway.But do look at ways to run giveaways that are engaging, memorable, and turn attendees into brand ambassadors.

Need a little inspiration? Run a raffle where the winner receives free lifetime product or access to a SaaS product. Give away free upgrades to those who sign-up at the event. Create a scavenger hunt with other companies where the winner receives an incredible grand prize.

Create experiences with your product giveaways to create lasting impressions while providing value. Product giveaways that provide real value mean more than a free t-shirt or pair of socks.

When planning for the next event or conference, transition to using inbound methodology. Attract customers by providing unique and valuable experiences that leave them wanting to learn more about your company and always remember this quote.

a�?Ia��ve learned that people will forget what you said, people will forget what you did, but people will never forget how you made them feel.a�?a�Sa��a�SMaya Angelou

Learn more about SummitSync.

Deciding how much to spend when shopping for a ring is a difficult task. Wea��ve heard every recommendation imaginable: Two to three montha��s salary, as much as you can afford, as little as possible, as much as possible. We wanted to break down all these theories to help you decide which fits you best.

The Skya��s the Limit

Ita��s easy to get carried away and spend wildly on an engagement ring. After all, the bigger the diamond, the more you love her right? Wrong. Diamonds are definitely an important piece of the pie, but there is such a thing as over the top. We dona��t believe a ten carat diamond ring would look nice on everyonea��s finger. Instead, if your budget is truly limitless, focus on factors other than size. A two, three or four carat engagement ring is large enough for most girls. Aim for for higher color, clarity and cut if your budget has room.

Conclusion: If the a�?skya�? is the limit for your budget, be sure to evenly focus on style, design, cut, clarity, carat and color. The ring should wow her and show her that youa��ve made a careful decision that suits her personality.

A�Diamonds are Worthless

Working within this industry, we see the appeal and value of diamonds. Diamonds have both tangible and intangible value. They are a symbols of love, status, commitment, loyalty, affection and basically everything else associated with a life-long relationship.

While we agree that buying diamonds purely for monetary gain or investment couldA�be risky, you can generally expect that the value of your diamond will rise over time.

Conclusion: Diamonds are not worthless. They have monetary and sentimental value. They are an accepted symbol of love and commitment. Luckily A�there are beautiful options available that will brighten her world within nearly any budget range.

Two to Three Monthsa�� Salary

Yes, this is a myth. Why did this become popular? Well, because it is the only suggestion that is accompanied by an actual dollar recommendation.

If you have a median level of income, carefully weigh what your average disposable income is in accordance with your lifestyle. If your lifestyle and budget do not allow you to spend two to three months salary on a ring, reassess your budget. Even within a smaller budget you can buy a gorgeous ring that she will love. You can always upgrade her ring when youa��ve saved enough at the next big milestone in your relationship.

If you decide that you can be more flexible with your disposable income, there still isna��t a hard and fast rule to determine your ring budget. As a general guide: the average diamond purchased for a proposal is H-color, SI1-clarity, and 0.90ct size. That will cost approximately $4,500. Use our on-site budgeting tool to understand just how much you should be spending.

The key is to determine which factors are most important to you (and her). If it is size, scale down on color, if it is clarity, pull down size. There absolutely needs to be a balance.

Conclusion: Spend what you can afford, no more, no less. You have to find that comfort zone that allows you to say you are satisfied and confident with your decision. You are buying a symbol of love, make sure you feel it!

Learn more about Four Mine.

2016 was anything but basic. The accelerated rise of indie players, sharing economy and adoption of tech-enabled lifestyles have defined the past twelve months. In 2016, Amazon started selling Kickstarter products, WeWork launched co-living apartments WeLive, Alexa learned over 5,000 skills and wearables got more fashionable. Across fashion, beauty, accessories, food, wellness, and home, consumer expectations are shifting and the retail landscape has changed from the bottom up. In 2017, and beyond, consumers will adopt connected products and disruptive models at an accelerated speed.

FROM SOMEONE WHO HAS WITNESSED COUNTLESS INTERVIEW-FAILS a�� AND FAILED A FEW HERSELF.

Authora��s Note: Hi, Ia��m one of the co-founders of Scouteda��a job matching platform for early career talent. I wear multiple hats here (disclaimer: blogger is usually not one of them), but my main role involves working with our partner companies. Ia��m constantly talking to our hiring managers to get feedback on the candidates theya��ve interviewed through our platform. A�Below, Ia��ve gathered my thoughts on common interview pitfalls. Check a�?em out, write a�?em down (cough, see #3), and get ready to crush your next set of interviews.

Interview Fail #1: Youa��re not prepared.

Grooan, youa��re saying. I thought this list was supposed to give me actual insider intel into the interview process. A�I know, I know. It sounds obvious, but we hear this complaint from hiring teams ALL THE TIME. A�So wea��re putting it first. A�Before every interview, some absolute musts:

1. Review the companya��s website, LinkedIn, social media profiles and Google them.
2. If possible, test out their product or service so you can provide feedback during the interview.
3. Often, you will be asked what makes the company unique in their space. A�This is a soft-ball. You should be able to answer this well and you should be able to explain why this company is different than similar ones in their field. A�(Hint: that means you know some of their competitors.)
4. Google the interviewer if you know them! A�Check them on Linkedin. A�It might even make you connect with the interviewer better!

Interview Fail #2: You have no clear passion for the company or job. A�

I cana��t count the number of clients who have said, a�?We really liked Corey, but he just didna��t seem that interested in the job.a�? Or, a�?Topanga would be awesome for the role, but she couldna��t articulate why she wanted the job or what she liked about our company.a�? A�FAIL. (Success, though, on winning our team bet to include a Boy Meets World reference in my blog article.)A�You are going to be asked what excites you about this opportunity and if you cana��t answer this question with genuine passion and conviction, you will fail to get the job. A�People care deeply about their work and want to see that you will care too.

Interview Fail #3: You dona��t have a pen and paper. A�

An interview is like a mock work day: people are trying to get a sense of what you would be like in their open role. A�You want to appear professional, and the first sign of reliability is having a pen and paper so you can take notes. Hopefully you will have questions and if you care about the answers, you should care enough to write down what the interviewers says. A�Even if you have a photographic memory like my childhood hero Cam Jansen, taking notes signals to the interviewer that you respect them and the time theya��ve taken out of their busy day to talk to you. A�And yea, we do mean a paper and pen. A�Technology is great but it is WAY more distracting in a job interview.

Interview Fail #4: You dona��t leverage your connections.

Internal referrals build social capital. If you know someone at the company, you should reach out before your interview. A�(Unless you stole their girlfriend in college or hazed the sh*t out of them during pledging. Then, perhaps lie low.) Ita��s a great opportunity to get insight into the company and opportunity. A�This is not about nepotism a�� ita��s much more practical than that. Ita��s hard to really get to know someone in the span of an interview or two, so if you have a guy saying a�?I know her, shea��s cool as icea�?, that can go a long way. A�And leta��s talk numbers, referrals are 7% of the average applicant pool but 40% of all hires! A�Plus, referrals get hired faster. A�Who doesna��t want those odds?

Interview Fail #5: You dona��t follow a�?best-in-classa�� interview principles.

When we were developing our initial product, a wise product manager helped me see the difference between a�?best in classa�� and a�?differentiatorsa��. A�A a�?best in class featurea�� is one that all your competitors have, but without it you wona��t be at the bar a�� A�like an easy to navigate website. (A a�?differentiatora�� is something that none of your competitors have, like on-demand resume feedback). A�When it comes to your job search, you want to make sure you exemplify all the standard a�?best in class featuresa�� to ensure youa��re a real contender. This means, write THANK YOU NOTES within 24 hours of your interview and always be polite and courteous to everyone you speak/interact with. Ita��s a small world out there and even if you dona��t get the job, you never know when your paths might cross again.

Interview Fail #6: You ask inappropriate questions too soon. A�

In a lot of ways, an interview is like a first date. Even if you really want to know if his maternal grandfather exhibited early-pattern baldness, you probably wait until at least Date 4 to pop that question. Same rules apply here: A�If you ask questions about compensation and work hours in the first interview, the company may think you care about the wrong things. A�These are important topics you should know about before deciding whether to accept an offer, but best to wait until you are going steady, if you know what I mean.

Interview Fail #7: You dona��t have any questions at all. 🙁

Back to that a�?passiona�? thing again a�� people want to see that you are genuinely interested in and excited about their opportunity. A�Ita��s OK to ask the same questions to different people a�� in fact it can be a good point of triangulation a�� but if you dona��t have any questions, most interviews will take that as a negative sign.

Interview Fail #8: You end an interview preemptively.

This also happens more often than you might think. Options have inherent value, so dona��t do anything that might prevent you from getting an offer. A�You owe it to yourself to get the opportunity to turn it down. A�ALSO, sometimes an interviewer will realize youa��d be a good fit for a different role, and will thus move you to that pipeline. So you should be open-minded, learn whatever you can about the opportunity and get the offer. A�Once you have it, then you SHOULD be selective about whether this is a good fit for you. Ita��s hard enough to get an interview, let alone a job. Dona��t be your worst enemy.

Thata��s all Ia��ve got for today! Hope it helps a�� and if youa��re eager to test your newly sharpened interview skills, come visit us at Scouted. Wea��ll hook you up.

Unbundled Advertiser Assets at Scale

Next to necessity, design is what motivates what we do and consume. It is evident in most decisions we make. Design is why you are wearing those clothes. Design is why you are carrying that bag or maybe no bag at all. And design is why you are reading this here.

CreativeA�is king, but when it comes to native campaigns, advertising creative is often viewed as cumbersome. Advertisers think they need to build custom executions for each individual site a�� or even worse, multiple executions for each site. In reality, the best programmatic native platforms can take a small number of unbundled assets to build rich, custom units on demand without any intervention on the advertisera��s part.

Programmatic Respects Design

In his inaugural Design in Tech Report at SXSW last year, former president of Rhode Island School of Design, John Maeda, spoke to the importance of combining, a�?business thinking with effective design and great engineering to achieve unflagging performance.a�? Focusing on the need for scalability of good design, Maeda stressed, a�?Design isna��t just about beauty; ita��s about market relevance and meaningful results.a�? In a world where publishers seem to have frequent redesigns, advertisers fear a stream of endless revisions for their in-feed creative deliverables. However, when buyers partner with native platforms that have a strong adherence to the design principles of the publishers they work with, creative isA�scalable and automated.

Making Design Scaleable for Native Ads

Before programmatic, many publishers had their own disparate approaches to native. They required advertisers to be very involved in creative executions, responding to a host of particular specs for each platform. This approach is incredibly time consuming. In turn this becomes the biggest impediment for building native programs. From adjusting for resolution to appropriately cropping a photo for the look and feel of a particular publisher, the day of a brand designer could be consumed by a seemingly endless list of tedious tasks. This amount of customization required by publishers can result in nearly endless back and forth revisions with advertisers. And ultimately, a campaign could take months to launch.

Today, advertisers see promise in using programmatic to bring automation and ad creative capabilities to native advertising, but they arena��t moving at full steam. Instead, theyA�are still apprehensive about whether or not native can truly be automated in a beautiful and contextually engaging manner. They worry that their exceptional designers will feel as though their talents are being squandered in an effort to execute simple, but many, design edits.

Publishers are also concerned. They fear that as their native units move toward real-time bidding and programmatic processes, they will lose control over what their product looks like and their integrity will suffer a blow. This is where, Maeda feels it importantA�a�?to connect the worlds of design, technology, and business together, so that people [do not continue to] see them as three different things.a�?

When advertisers partner with technology platforms to create native advertising that aligns itself with the look, content, and feel of the publishera��s environment, the worlds of design, technology, and business, come together to create scalable solutions. Advertisers can continue dedicating their creative resources effectively, while publishersa�� continue to deliver quality content to their audience.

The Technology Connecting Design and Business.

Using a proprietary version of computer vision, a method of A�acquiring, processing, analyzing, and understanding visual assets, TripleLift decomposes images based on the most important elements and then renders them differently on thousands of sites resulting in a custom experience with no intervention from both advertisers and publishers. For instance, if one publisher has a spec sheet that does not meet standard size requirements, sophisticated computer vision technology can crop the photo so that valuable pieces like product imagery and logo remain seamlessly.

Getting Back to Quality Creative

The smartphone revolution has significantly increased consumer demand for quality design. Consumers expect to hit the homepage of their favorite publisher and browse through an interface that is both beautiful and intuitive. Publishers are meeting this demand by delivering quality content and expect the same care with in-feed native advertising. Creating and scaling custom, high quality, native advertising across platforms can be challenging and weigh on an advertisera��s creative teams. The room for error is non existent and in a non-automated native campaign, an advertisera��s must take great care to comply with the different design principles of publishers across all devices. When advertisers partner with a native technology platform like TripleLift, one piece of creative can be scaled to over a thousand premium publishers.

By leveraging pioneering computer vision technology, advertisers can relieve their creative teams of the burdens that come with creating traditional native advertising, and allow them to focus more on the creative strategy that compelled them to enter the world of advertising in the first place.

Learn more about TripleLift.

Sources:

http://www.wired.com/2015/03/take-expert-design-important-ever/

http://www.kpcb.com/blog/design-in-tech-report-2015

http://www.slideshare.net/kleinerperkins/design-in-tech-report-2016/3-Reports_from_Gartner_Accenture_Deloitte