Identifying gibberish content in resources

A non-transitory computer-readable storage medium having instructions stored thereon that, when executed by data processing apparatus, cause the data processing apparatus to perform operations comprising:

• obtaining a query index of a plurality of different queries submitted by a plurality of users to a search engine during a given time period, each query of the plurality of queries being keyed to one or more terms found in the respective query such that the query index identifies, for each key, one or more queries containing the key;
• obtaining a resource, the resource including text contents;
• calculating a language model score for the resource including applying a language model to the text content of the resource;
• calculating a query stuffing score for the resource based on, at least, for a plurality of most frequently occurring terms in the text content of the resource that each match a respective key in the query index, a ratio of a count of queries in the index keyed to the term that appear in one or more phrases in the text content to a total count of queries in the index keyed to the term;
• and
• generating the query stuffing score based on an average of the calculated ratios;
• calculating a gibberish score for the resource based on the language model score and the query stuffing score;
• and
• determining whether to modify a ranking score of the resource based on the calculated gibberish score.

A system comprising:

• a memory and a processor configured to perform operations including:
• obtaining a query index of a plurality of different queries submitted by a plurality of users to a search engine during a given time period, each query of the plurality of queries being keyed to one or more terms found in the respective query such that the query index identifies, for each key, one or more queries containing the key;
• obtaining a resource, the resource including text contents;
• calculating a language model score for the resource including applying a language model to the text content of the resource;
• calculating a query stuffing score for the resource based on, at least, for a plurality of most frequently occurring terms in the text content of the resource that each match a respective key in the query index, a ratio of a count of queries in the index keyed to the term that appear in one or more phrases in the text content to a total count of queries in the index keyed to the term;
• and
• generating the query stuffing score based on an average of the calculated ratios;
• calculating a gibberish score for the resource based on the language model score and the query stuffing score;
• and
• determining whether to modify a ranking score of the resource based on the calculated gibberish score.