The explanation is incomplete or not explained properly. The below explaination is taken from Core Java Volume 1:
A linked hash map can alternatively use access order, not insertion order, to iterate through the map entries. Every time you call get or put, the affected entry is removed from its current position and placed at the end of the linked list of entries. (Only the position in the linked list of entries is affected, not the hash table bucket. An entry always stays in the bucket that corresponds to the hash code of the key.) To construct such a hash map, call
LinkedHashMap<K, V>(initialCapacity, loadFactor, true)
Access order is useful for implementing a “least recently used” discipline for a cache. For example, you may want to keep frequently accessed entries in memory and read less frequently accessed objects from a database. When you don’t find an entry in the table, and the table is already pretty full, you can get an iterator into the table and remove the first few elements that it enumerates. Those entries were the least recently used ones.
You can even automate that process. Form a subclass of LinkedHashMap and override the method
protected boolean removeEldestEntry(Map.Entry<K, V> eldest)
Adding a new entry then causes the eldest entry to be removed whenever your method returns true. For example, the following cache is kept at a size of at most 100 elements:
Map<K, V> cache = new
};
Alternatively, you can consider the eldest entry to decide whether to remove it. For example, you may want to check a time stamp stored with the entry.
