In one embodiment, a method includes receiving a text query comprising one or more n-grams inputted by a first user, identifying one or more primary entities matching one or more n-grams of the received text query, and identifying, for each of the identified primary entities, one or more related entities based on one or more related-entity indexes associated with the primary entity. The method also includes accessing, for each identified related entity, one or more posts authored by the identified related entity, each accessed post matching all of the n-grams of the received text query, calculating a score for each of the accessed posts, and sending to the first user one or more search results corresponding to one or more of the accessed posts having scores higher than a threshold score, respectively.