Google Patent Attacks Reverse Engineering of Local Search Listings

The title from a Google patent reached out and grabbed me as I was skimming through Google’s patents. It has the kind of title that captures your attention, as a weapon in the war that Google wages against people who might try to spam the search engine.

The title for the patent is Reverse engineering circumvention of spam detection algorithms. The context is local search, where some business owners might be striving to show up in results in places where they don’t actually have a business location, or where heavy competition might convince them that having additional or better entries in Google Maps is going to help their business.

The result of such efforts might be for their local listings to disappear completely from Google Maps results. The category Google seems to have placed such listings under is “Fake Business Spam.”

Google spam score flow chart from patent

Google has filed and been granted a number of patents in a category I’ve called Web Spam.

I like finding those, reading about them, and making sure that they don’t inadvertently target sites that they shouldn’t. A false positive, or a site mistakenly positively identified as spamming a search engine, may mean more than just a site getting demoted in rankings.

It could mean that sites might just disappear from rankings completely.

Rank-Modifying Spammers Patent

One patent, filed a couple of years back, took sites that Google was suspicious of using spam to improve rankings and instead of increasing their ranks, decreased their rankings, or kept them the same, or let rankings fluctuate randomly.

This kind of incorrect ranking was intended to get a spammer to reverse changes they may have made to a site, or roll some of them back.

If the changes were reversed by (possibly the same person), Google might decide that the changes were done by a spammer, especially if new changes were even more spammy (unnatural links, keyword stuffing, etc.). I titled my post about that patent, The Google Rank-Modifying Spammers Patent.

The post and patent inspired a few discussions and a few questions to the Google Spam fighting team. Matt Cutts released a video sometime after, warning people that, “Just because we have a patent on something doesn’t mean we are currently using it.

Fake Business Spam

Google local business listings provide data identifying a business and its location both in the world and on the web. It targets helping people find businesses of certain types, close to specific locations. People do create bogus listings, attached to very real phone numbers to generate business leads. Often many of such businesses don’t necessarily even need nearby offices to conduct the kind of business that they perform. As the patent warns us:

The customer may be defrauded by contacting or visiting an entity believed to be at a particular location only to learn that the business is actually operating from a completely different location. Such fraudulent marketing tactics are commonly referred to as “fake business spam”.

We’re next told that search engines try to identify such spammers, who sometimes experiment with search listing submissions to see if they can manipulate search results. They might do this by generating a spam score for business listings based upon a number of factors and then adding some random numbers to that score so that it isn’t easy for spammers to reverse engineer those scores. This seems to be an inspiration for making such changes:

The amount of noise that is added is sufficient to affect spammer’s attempts to reverse engineer spam-identification algorithms, but small enough so that the search experience of end users is minimally affected because the ranking of legitimate listings is unaffected.

Reverse engineering circumvention of spam detection algorithms
Invented by Douglas Richard Grundman
Assigned to Google
US Patent 8,612,436
Granted December 17, 2013
Filed September 27, 2011

Abstract

A spam score is assigned to a business listing when the listing is received at a search entity. A noise function is added to the spam score such that the spam score is varied.

In the event that the spam score is greater than a first threshold, the listing is identified as fraudulent and the listing is not included in (or is removed from) the group of searchable business listings. In the event that the spam score is greater than a second threshold that is less than the first threshold, the listing may be flagged for inspection.

The addition of the noise to the spam scores prevents potential spammers from reverse engineering the spam detecting algorithm such that more listings that are submitted to the search entity may be identified as fraudulent and not included in the group of searchable listings.

Local Business Listings Spam Scores

The patent lists some of the things it looks at in generating an initial spam score, such as:

  • The geographic density of businesses in the same category
  • Repeated identifying information in different listings
  • Ratios of common terms in the business listing title to total words in the title
  • Any other number of known methods for determining whether a listing is fraudulent

I didn’t write this post to help people who might be spamming business search listings, but rather because I know how important search listings can be to legitimate businesses of all types that rely to a degree upon Google local listings for the success of their businesses.

The patent does provide more details on the process behind how they use a spam score and how random amounts of noise might be added. If you’re interested, I’d suggest digging into the patent a bit more.

Share

13 thoughts on “Google Patent Attacks Reverse Engineering of Local Search Listings”

  1. “The geographic density of businesses in the same category”

    Seems that this metric could be interpreted in multiple ways:

    1. A new listing is suspected as fake because it resides on the same block as four other competitors.

    2. A new listing is suspected as fake because there are no competitors in the area. Therefore an existing market is not present for the product/service.

    Either one looks like a poor candidate for detecting fakes.

  2. Hi Daniel,

    I can see the use of that metric as worthy of consideration. The density of businesses in the same category gains significance when a business resides on the same block as 50 competitors, or even when a business such as a locksmith suddenly appears in new locations in every town in a block of 5 contiguous counties. Both raise red flags that suggest further investigation.

  3. But specifically how is density a factor? New businesses popping up on every corner is suspicious irregardless of the business category’s density in those areas. And a singular new business listing in a densely competitive area seems like a very natural occurrence. You want to open doors where the market is.

  4. Hi Daniel,

    Remember, I didn’t write the patent. I’m just writing about it.

    What kinds of signals might you think would be potential indications of spam listings to people at Google?

  5. I remember ;-)

    I am primarily trying to determine if I am interrupting “geographic density” as intended. That could refer to the density of all of a *specific* business’s locations, or of the density of all of one *type* of business in a location.

    This is one area of spam that Google’s number crunching horsepower won’t take them very far. They need more frequently updated street maps and social data which they don’t have.

  6. Hi Daniel

    There are times when I think Google might be purposefully a little misleading, and this patent had that specifically as an intent.

    I have seen Starbucks across the street from each other in some cities, but it would be kind of strange to see 10 pizza places in a row.

    There are some types of businesses that seem to find addresses clustered around places like a Mailboxes etc., or a registered agent, and those kinds of things may be a sign that an address may not be the actual address that they conduct business at. It’s one of the difficulties of trying to create an actual index of the world without having shoes on the street. Joe mentions in a followup post to this one Googles acquisition of Skybox, which is something I wrote about too:

    Google’s Skybox Eye-in-the-Sky Acquisition

    Would Google use their satellites to check out businesses listed in Google Maps? They been supposedly using Recaptcha to supposedly check upon street numbers for buildings in Streetview, to help better match up addresses and locations for businesses.

  7. It’s great to hear that Google is coming up with new measures to fight spammers, but what if innocent people get caught in this new algorithm? Like Daniel said, just because a new business opens up in a densely occupied area doesn’t mean its a spammer.

  8. Hi David,

    The patent isn’t saying that it would automatically consider a business listing to be spam if it opens up in a densely occupied area, but instead says that it would look at “The geographic density of businesses in the same category” as a signal that it might look at to try to understand if a site is spam.

    Of course new businesses tend to open up in areas where there are other businesses. Some chain restaurants, for example, look for locations of other chain restaurants to open up near.

    But, a tow truck company opening and listing its address on the same block as 50 other tow truck companies is ridiculous and probably a sign of spam. Chances are that with that many tow companies in the space block, most of them are mail drops only, which is something that Google is trying to avoid completely. If you say your business is located there, but it really isn’t, then your listing is considered by Google to be Fake business spam, and the only reason why that listing might exist is to try to show potential customers that you have a local presence, even though you really don’t.

  9. Wow, Google is pulling out all of the stops to try to get people to fake out the system. They are certainly getting smarter with detecting spam and websites gaming its index.

  10. Very nice Article Bill, thanks for sharing such informative post with us. Google patent is used to describe how Google will randomly move the search engine ranking of a website when their are some rank-modifying spamming techniques in use.
    Anyone who is intentionally doing something other than “creating great content” for higher SERPs is a spammer .

    Definitely agree that certain spammy tactics still prevail in certain places.

    Good work

  11. In itself this article and discussion raises many interesting points. I’d like to throw another one into the mix. My experience is that Google rolls out any new “system” for catching spammers across all geographic regions but its own data is very poor in some areas. Working in the U.K., this can be quite common in certain regions.

    When Google hasn’t mapped an area properly and doesn’t offer StreetView it can create massive confusion amongst business owners. They just don’t “get” why Google doesn’t let them enter their own data in a format they understand. The result is that they either attempt to add data that Google thinks is wrong (even though it reflects the on-the-ground reality) OR some of them adopt Google’s errors and work with them. This muddles the data still further. In the chaos that follows, it’s not surprising that businesses inadvertently send off spam signals when they repeatedly fiddle with their business listings to try to get on the map, in the right place, with the right business category.

    I’d love to see spam-free search but I’d also like to find a way to let Google know when their data is wrong! Mapmaker is great for places that are already well mapped and where there are lots of data sources to call on to justify your edits but it’s not much help otherwise.

  12. Hi Kate,

    My experiences with Map Maker have been hit or miss. It’s a little like them (the map maker Google employees) not trusting you very much, which they probably don’t. When you tell them that a business has been closed for 2+ years, and they don’t trust you enough to remove it from Google Maps until you send them 10+ links to websites, include local newspaper reports, that state when the business has closed, they start getting a little more trusting.

    A post I did a couple of years ago, http://www.seobythesea.com/2011/10/gps-to-correct-google-maps-and-driving-directions-as-a-local-search-ranking-factor/, tells us about a patent which suggests that if you report something to map maker on a mobile device, from the location where the business was at, they trust that report a little more, too.

    They probably need more feet in the street, or to get their street car drivers to get out of the cars and try some doors, or get more satellites up in the air.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>