This article originally appeared on LinkedIn.com.

Rating systems launched much of today’s online commerce. Innovative, brilliant and transformational, the early ratings systems helped build entire industries, from ride sharing to e-commerce. They created pools of trustworthy service providers and became a source of insight into the reliability of products and merchants, generating information that became the primary competitive advantage of many companies.

Over time, however, these once-pioneering ideas have eroded in value. Grade inflation has crept in, making it harder to decipher what’s truly great (4.6 on Uber’s 5-star rating scheme sounds strong, but is actually so low it can lead to removal of a driver from the platform). Fakery is an issue as well, with experts now contending that a third of all online reviews are bogus, either the intentional result of competitors’ bad actions or of click-farms and other purveyors of insincere praise.

Companies dependent on useful reviews are in urgent need of innovation that upgrades their reliability and relevance. Without it, rapid obsolescence is a real risk.

I was impressed recently by online travel site TripAdvisor’s determination to take on this challenge. The site, which depends on user reviews of hotels, restaurants and other visitor sites, fights vigorously to maintain the quality of those reviews, and takes aggressive action to protect them. In the past three years TripAdvisor has rooted out and stopped 60 companies that had been posting fake reviews.

They take this issue quite seriously. In addition to the standard steps of blocking users and penalizing businesses that post fraudulent reviews, TripAdvisor encourages law enforcement action and works with prosecutors pursuing cases of fraud.

In the most dramatic example to date, the company announced earlier this fall the nine-month prison sentence of a user in Italy by the Criminal Court of Lecce, the conclusion of a case based in part on TripAdvisor’s own sleuthing. The defendant’s company, PromoSalento, sold fake reviews to local hospitality businesses.

What TripAdvisor called a “landmark ruling for the Internet” started when several hospitality companies alerted the site that they had been approached by PromoSalento. TripAdvisor then launched an internal investigation, and after its fraud detection programs identified and blocked or removed more than 1,000 reviews posted by PromoSalento, it punished hundreds of suspected clients by demoting them in its rankings. In the most egregious cases, the company will also add a red badge to a property’s review, an alert that the business has tried to manipulate reviews.

Several clients of PromoSalento agreed to cooperate with TripAdvisor’s investigation, sharing confirmation of payments, bank transactions and service receipts. TripAdvisor shared this evidence with the Italian Postal and Communications Police, and the company’s Italian legal counsel also provided support.  

I applaud the company’s efforts and continue to think about features that might help protect the veracity of reviews and make them more meaningful.

One idea is to create a way for users of a site to boost reviews they have found especially relevant and helpful, creating a set of favored reviewers for each customer. This group would then receive heavier weighting (and elevated visibility of comments) anytime that user is on the site. It’s customization akin to Netflix suggesting the movies you will most enjoy by matching you to other viewers with similar tastes.

The last time I wrote about the pressures reviews are under, the piece launched an exciting debate among readers. Commenters suggested that technologies like natural language understanding could draw insight from many user comments in a way that creates reviews that are more broadly accurate. One reader suggested that a better system would be simpler, binary—either something worked or it didn’t. Another shared their experience that the pool of reviewers expands, and the quality of reviews improves, when you time the request for the review well. Others advocated that reviews be curated by an expert from the community, and that true experts be given a higher profile in reviews, as a few sites currently do.

It was an impressive outpouring of creative thinking, but this problem isn’t yet solved. So how would you upgrade the ratings and review systems you use today?