S1: obtaining Mi (i+1)^th-hop candidate documents based on a retrieval text and Ki i^th-hop candidate documents, wherein i, Ki, and Mi are positive integers, and wherein Ki is less than or equal to Mi;

S2: obtaining a first score for each of the Mi(i+1)^th-hop candidate documents, wherein the first score represents a relevance degree between a corresponding candidate document and the retrieval text query;

S3: obtaining, for a candidate document Pjy(i+1) in the Mi (i+1)^th-hop candidate documents and based on the first score of the candidate document Pjy(i+1) and a first probability of a path L, a second probability of a first path corresponding to the candidate document Pjy(i+1), wherein the path L starts with a first-hop candidate document and ends with a candidate document Pji in the Ki i^th-hop candidate documents, wherein the candidate document Pji is a j^th candidate document in the Ki i^th-hop candidate documents, wherein the candidate document Pjy(i+1) is a y^th candidate document that is in the Mi (i+1)^th-hop candidate documents and that is based on the candidate document Pji, and wherein both j and y are positive integers;

S4: obtaining K(i+1) (i+1)^th-hop candidate documents based on third probabilities of second paths corresponding to the Mi(i+1)^th-hop candidate documents, wherein the K(i+1) (i+1)^th-hop candidate documents are in the Mi (i+1)^th-hop candidate documents, wherein the third probabilities of the second paths of the K(i+1) (i+1)^th-hop candidate documents are higher than the second probability of the first path of a remaining candidate document in the Mi (i+1)^th-hop candidate documents, and wherein K(i+1) is a positive integer less than or equal Mi; and

S5: obtaining, based on the K(i+1) (i+1)^th-hop candidate documents, a retrieval result corresponding to the retrieval text query.