searching in the query statement set to obtain at least one first query statement in the query statement set, the at least one first query statement each satisfying a first sentence structure condition corresponding to a first target entity word; wherein, in response to determining that a first noun in a query statement is the first target entity word and there is only one another noun following the first target entity word in the query statement, determining that the query statement is a first query statement satisfying the first sentence structure condition corresponding to the first target entity word, determining the only one another noun following the first target entity word in the first query statement as a second level entity word, and determining the second level entity word as a child node of the first target entity word;

mining in the at least one first query statement, to obtain at least one second level entity word corresponding to the first target entity word;

merging the first target entity word and each second level entity word in the at least one second level entity word corresponding to the first target entity word to obtain a second target entity word, and searching in the query statement set to obtain a second query statement in the query statement set, the second query statement containing the second target entity word and satisfying a second sentence structure condition corresponding to the second target entity word; and

mining in the second query statement, to obtain a third level entity word corresponding to the second target entity word, and determining the third level entity word as a child node of the second level entity word,

wherein, in response to determining that a first noun in a query statement in the query statement set is the second target entity word and there is only one verb following the second target entity word, determining that the query statement is a second query statement satisfying the second sentence structure condition corresponding to the second target entity word, and determining the only one verb following the second target entity word as the third level entity word; and

wherein the method further comprises,

extracting an entity word from associated entity words corresponding to the first target entity word as a current associated entity word corresponding to the first target entity word;

counting an access frequency or a number of access times of the first target entity word and the current associated entity word corresponding to the first target entity word in a predetermined time period;

in response to the access frequency being greater than a preset frequency threshold or the number of access times being greater than a preset number of times threshold, determining that the current associated entity word corresponding to the first target entity word is valid;

repeating the extracting, counting and determining operation until whether each of the associated entity words corresponding to the first target entity word is valid is verified.